# forck.live — 3–9 August 2026

> AI lab announcements, news reports and podcasts.

forck.live follows what companies announce about their own products. Press may
corroborate an official story; it never leads and never appears as an item of its
own. A missing date stays missing. A title is the publisher's words — where a
company does not publish in English, an English rendering is listed separately
as `Title in English` and named as this site's translation, never as theirs.

This file is one issue of the weekly archive: every confirmed
first-party announcement of ISO week 2026-W32, 3–9 August 2026, newest
first. The window never moves, so this URL always names the same seven days —
but a post published inside them and detected later is added when it arrives,
so a closed week can still gain entries.

Every entry is one first-party announcement, and carries the same eight fields
in this order before any others:

    Company · Announced · Category · Coverage · Announcement · Group · Source · Record

`Source` is the company's own page the announcement was traced to and `Record`
is this site's page for it. Both are safe to cite.

`Coverage` is a number of outlets, or `not counted` — which means this record
reached no count of its own, never that nobody wrote about it.

`Category` is this tracker's own analysis, and is authoritative wherever it is
stated. `not stated` means either that nothing has analysed the post yet or
that the analysis placed it outside this taxonomy; neither is a verdict about
the announcement.

`Announcement` is whether the post is news of something the company did, rather
than a how-to, a customer story, a changelog listing or a staffing note.

`Group` is which of this record's groups the entry falls in, carried on the
entry itself so it holds where the groups are headings and where they are not.
One of: `models`, `covered`, `announcements`, `routine`.

Anything after `Record` is held detail that only some announcements carry.

- This week's page: https://forck.live/week/2026-W32
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### NVIDIA — Firebird Launches CIS Region’s Largest AI Factory in Armenia

- Company: NVIDIA (nvidia.com)
- Announced: 2026-08-08T10:24:54+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://blogs.nvidia.com/blog/firebird-ai-factory-armenia-blackwell-rubin-dsx/
- Record: https://forck.live/items/3688-firebird-launches-cis-region-s-largest-ai-factory-in-armenia
- Subject: AI platform

Firebird launched the CIS region's largest AI factory in Armenia, powered by NVIDIA accelerated computing and Dell Technologies infrastructure. The factory will deploy over 70,000 NVIDIA Rubin and Blackwell GPUs by end of 2027. NVIDIA intends to invest in Firebird.

### GitHub — GitHub Copilot weekly releases — August 3

- Company: GitHub (github.com)
- Announced: 2026-08-07T23:16:02+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-07-github-copilot-weekly-releases-august-3
- Record: https://forck.live/items/3680-github-copilot-weekly-releases-august-3
- Subject: Copilot

GitHub Copilot Auto now displays the model used for each request along with AI credit and cache details.

### GitHub — Copilot impact dashboard adds a return on investment section

- Company: GitHub (github.com)
- Announced: 2026-08-07T21:05:37+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-07-copilot-impact-dashboard-adds-a-return-on-investment-section
- Record: https://forck.live/items/3682-copilot-impact-dashboard-adds-a-return-on-investment-section
- Subject: Copilot

GitHub Copilot's impact dashboard now includes a return on investment section that compares spending and pull request output across developer adoption phases, with a salary selector for modeling.

### GitHub — Copilot code review effort levels are generally available

- Company: GitHub (github.com)
- Announced: 2026-08-07T20:49:26+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-07-copilot-code-review-effort-levels-are-generally-available
- Record: https://forck.live/items/3683-copilot-code-review-effort-levels-are-generally-available
- Subject: Copilot

GitHub Copilot code review now offers two generally available effort levels, Lite and Balanced, replacing the preview names Low and Medium. Lite is for straightforward changes, while Balanced uses a higher-reasoning model for deeper analysis.

### GitHub — Copilot usage metrics API adds agent app activity

- Company: GitHub (github.com)
- Announced: 2026-08-07T18:20:47+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-07-copilot-usage-metrics-api-adds-agent-app-activity
- Record: https://forck.live/items/3670-copilot-usage-metrics-api-adds-agent-app-activity
- Subject: Copilot

The Copilot usage metrics API now reports activity from third-party agent apps, broken out by individual agent, allowing teams to track usage per agent.

### Amazon — How Cohere Health digitizes clinical policies using Amazon Bedrock AgentCore

- Company: Amazon (amazon.com)
- Announced: 2026-08-07T16:26:58+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/how-cohere-health-digitizes-clinical-policies-using-amazon-bedrock-agentcore/
- Record: https://forck.live/items/3658-how-cohere-health-digitizes-clinical-policies-using-amazon-bedrock-agentcore
- Subject: Bedrock / Nova

Cohere Health built Cohere Policy Studio using Amazon Bedrock AgentCore to digitize clinical policies, enabling multi-tenant architecture and managed agent runtime for prior authorization automation.

### Amazon — How TReNDS automates root-cause analysis with Amazon Bedrock

- Company: Amazon (amazon.com)
- Announced: 2026-08-07T16:22:50+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/how-trends-automates-root-cause-analysis-with-amazon-bedrock/
- Record: https://forck.live/items/3659-how-trends-automates-root-cause-analysis-with-amazon-bedrock
- Subject: Bedrock / Nova

Amazon Bedrock is used as the core reasoning engine in a system that automates root-cause analysis of application errors by combining CloudWatch logs, Lambda, and the Strands Agents SDK.

### Amazon — Determining playoff clinching scenarios in the NHL using constraint programming

- Company: Amazon (amazon.com)
- Announced: 2026-08-07T16:21:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/determining-playoff-clinching-scenarios-in-the-nhl-using-constraint-programming/
- Record: https://forck.live/items/3660-determining-playoff-clinching-scenarios-in-the-nhl-using-constraint-programming
- Subject: Bedrock / Nova

AWS Generative AI Innovation Center created an automated system using constraint programming and custom tree search to determine NHL playoff clinching scenarios, validated against official NHL results.

### OpenAI — Responding to the next frontier of critical cyber capabilities

- Company: OpenAI (openai.com)
- Announced: 2026-08-07T15:20:00+00:00
- Category: safety-policy-update
- Coverage: 21 outlets
- Announcement: yes
- Group: covered
- Source: https://openai.com/index/responding-next-frontier-critical-cyber-capabilities
- Record: https://forck.live/items/3656-responding-to-the-next-frontier-of-critical-cyber-capabilities
- Subject: GPT / ChatGPT / API

OpenAI is sharing preliminary cybersecurity evaluations for Astra and the steps they are taking to strengthen safeguards and security controls.

### GitHub — GitHub Code Quality no longer adds Copilot as a reviewer

- Company: GitHub (github.com)
- Announced: 2026-08-07T15:07:25+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-07-github-code-quality-no-longer-adds-copilot-as-a-reviewer
- Record: https://forck.live/items/3661-github-code-quality-no-longer-adds-copilot-as-a-reviewer
- Subject: Copilot

GitHub Code Quality no longer automatically creates a ruleset that requests Copilot code review on pull requests; existing rulesets have been disabled, but users can manually re-enable the feature.

### Krea — Bending the Kubernetes scheduler

- Company: Krea (krea.ai)
- Announced: 2026-08-07T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.krea.ai/blog/bending-the-kubernetes-scheduler
- Record: https://forck.live/items/15099-bending-the-kubernetes-scheduler
- Subject: Krea 2 / AI creative platform

Krea describes how it uses dynamic taints driven by a PromQL signal, the Descheduler, and Kueue to keep GPU workloads in the cluster and spill only when capacity is exhausted, without modifying the default Kubernetes scheduler.

### Google DeepMind — See what 5 builders are making with Gemini Omni

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-08-07T14:00:00+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-omni-builders/
- Record: https://forck.live/items/3657-see-what-5-builders-are-making-with-gemini-omni
- Subject: Gemini

Google DeepMind highlights five builders using Gemini Omni to create and edit videos through natural conversation, demonstrating capabilities like changing camera angles, transforming scenes, animating sketches, applying styles, and visualizing concepts.

### OpenAI — How HSP GRUPPE builds AI capabilities for tax advisory

- Company: OpenAI (openai.com)
- Announced: 2026-08-07T09:00:00+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/hsp-gruppe
- Record: https://forck.live/items/3654-how-hsp-gruppe-builds-ai-capabilities-for-tax-advisory
- Subject: GPT / ChatGPT / API

HSP GRUPPE uses ChatGPT Enterprise to boost productivity, improve work quality, and create more capacity for tax advisory and client service.

### GitHub — MCP allowlists in enterprise managed settings

- Company: GitHub (github.com)
- Announced: 2026-08-07T00:11:19+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-06-mcp-allowlists-in-enterprise-managed-settings
- Record: https://forck.live/items/3673-mcp-allowlists-in-enterprise-managed-settings
- Subject: Copilot

Enterprise owners can now centrally control which MCP servers GitHub Copilot clients are allowed to run by using the new allowedMcpServers and deniedMcpServers keys in enterprise managed settings.

### Supabase — Supabase is now a connector on Perplexity Computer

- Company: Supabase (supabase.com)
- Announced: 2026-08-07
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/supabase-is-now-a-connector-on-perplexity-computer
- Record: https://forck.live/items/21273-supabase-is-now-a-connector-on-perplexity-computer
- Subject: Supabase

Supabase is now available as a connector on Perplexity Computer, allowing the agent to read from and write to Postgres tables, keep state across runs, and perform tasks such as querying production data, debugging support tickets, testing Edge Functions, and scheduling recurring queries.

### Ai2 — TutorMoments: Do AI tutors know when to help and when to hold back?

- Company: Ai2 (allenai.org)
- Announced: 2026-08-07
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://allenai.org/blog/tutormoments
- Record: https://forck.live/items/18540-tutormoments-do-ai-tutors-know-when-to-help-and-when-to-hold-back
- Subject: Olmo / Molmo / Asta

Ai2 introduced TutorMoments, a framework and dataset for evaluating whether large language models can appropriately decide when to scaffold or push for rigor in math tutoring. Based on 462 real tutoring transcripts with over 1,500 teacher-annotated key moments, the evaluation found that models tend to over-help and rarely push students to deeper thinking, even when the trade-off is spelled out in the prompt. Ai2 released the de-identified transcripts, replay pipeline code, and model replays for reproducibility.

### Sarvam AI — Everything we announced at Sarvam Epoch

- Company: Sarvam AI (sarvam.ai)
- Announced: 2026-08-07
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.sarvam.ai/epoch/summary/
- Record: https://forck.live/items/18385-everything-we-announced-at-sarvam-epoch
- Subject: Sarvam / Bulbul / Saaras

Last week, we held the first edition of Sarvam Epoch. It was energising to see the community come together at Epoch with such a strong sense of possibility.

### Amp — Size the Orbs of Production!

- Company: Amp (ampcode.com)
- Announced: 2026-08-07
- Category: pricing-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/size-the-orbs-of-production
- Record: https://forck.live/items/7717-size-the-orbs-of-production
- Subject: Amp

Amp announced a 20% price cut on orbs, introduced a new a1.medium size, reduced auto-pause time from 15 to 5 minutes, sped up startup times, and allowed per-thread orb size selection. The new pricing tier includes sizes from a1.tiny at $0.08/hour to a1.xxlarge at $1.32/hour.

### xAI — Imagine Image 2.0

- Company: xAI (x.ai)
- Announced: 2026-08-07
- Category: model-update
- Coverage: 1 outlet
- Announcement: yes
- Group: models
- Source: https://x.ai/news/grok-imagine-image-2
- Record: https://forck.live/items/4349-imagine-image-2-0
- Subject: Grok

xAI announces the general availability of Imagine Image 2.0, a new quality mode for image generation and editing on grok.com and mobile apps, with features like magic wand editing, segmentation, background removal, and multi-ref editing. It ranks second in the world in both text-to-image generation and image editing according to Arena leaderboards. The model is also available via API as grok-imagine-image-2.0.

### Anthropic — Improving Fable 5's biology safeguards

- Company: Anthropic (anthropic.com)
- Announced: 2026-08-07
- Category: product-launch
- Coverage: 8 outlets
- Announcement: yes
- Group: covered
- Source: https://www.anthropic.com/news/improving-fable-5-s-biology-safeguards
- Record: https://forck.live/items/3653-improving-fable-5-s-biology-safeguards
- Subject: Claude

Anthropic announces updates to Claude Fable 5's biology safeguards that reduce false positives, cutting biology-related fallbacks by about 85%. This allows Fable 5 to assist with a wider range of biology tasks, including everyday health and educational questions, while still falling back to Opus 5 for dual-use requests.

### AMD — AMD Acquires Taalas to Advance Compute Solutions for Rapidly Growing AI Inference Market

- Company: AMD (amd.com)
- Announced: 2026-08-06T20:05:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ir.amd.com/news-events/press-releases/detail/1296/amd-acquires-taalas-to-advance-compute-solutions-for-rapidly-growing-ai-inference-market
- Record: https://forck.live/items/15033-amd-acquires-taalas-to-advance-compute-solutions-for-rapidly-growing-ai
- Subject: AMD Instinct / ROCm

AMD announced a definitive agreement to acquire Taalas, a company specializing in AI inference silicon. The acquisition aims to strengthen AMD's AI roadmap with differentiated inference technology and engineering expertise, with plans to integrate Taalas' technology into AMD's accelerator roadmap and develop system-level solutions alongside AMD Instinct GPUs.

### Amazon — Securing AI agents with temporal policies in Amazon Bedrock AgentCore

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T18:57:55+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/securing-ai-agents-with-temporal-policies-in-amazon-bedrock-agentcore/
- Record: https://forck.live/items/3627-securing-ai-agents-with-temporal-policies-in-amazon-bedrock-agentcore
- Subject: Bedrock / Nova

Amazon announced temporal policies for Amazon Bedrock AgentCore, which enforce stateful authorization rules by evaluating each request against the agent's session history. The policies run at the gateway perimeter and cannot be bypassed by the agent.

### Amazon — Configure rate limits for AI traffic on AgentCore gateway

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T17:50:42+00:00
- Category: infrastructure-release
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/configure-rate-limits-for-ai-traffic-on-agentcore-gateway/
- Record: https://forck.live/items/3628-configure-rate-limits-for-ai-traffic-on-agentcore-gateway
- Subject: Bedrock / Nova

Amazon Bedrock AgentCore gateway now supports rate limiting, allowing per-user control over traffic to tools, inference models, and agents through request, token, and connection limits.

### GitHub — Kimi K3 is now available in GitHub Copilot

- Company: GitHub (github.com)
- Announced: 2026-08-06T17:27:25+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-06-kimi-k3-is-now-available-in-github-copilot
- Record: https://forck.live/items/3635-kimi-k3-is-now-available-in-github-copilot
- Subject: Copilot

GitHub announced general availability of Kimi K3, an open-weight model, in GitHub Copilot. The model offers frontier-level agentic coding abilities at a pricing of $3 per 1M input tokens, $15 per 1M output tokens, and $0.30 per 1M cached input tokens. It is hosted on Fireworks AI and available across multiple Copilot plans and tools. The rollout is temporarily paused due to a GitHub Actions incident but will resume.

### Amazon — Control agent behaviors and cost beyond a single action: new capabilities in Amazon Bedrock AgentCore

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T16:43:19+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/control-agent-behaviors-and-cost-beyond-a-single-action-new-capabilities-in-amazon-bedrock-agentcore/
- Record: https://forck.live/items/3629-control-agent-behaviors-and-cost-beyond-a-single-action-new-capabilities-in
- Subject: Bedrock / Nova

Amazon Bedrock AgentCore introduces temporal policies powered by Dogwood (a new open source policy language) and rate limiting in the gateway, enabling teams to control agent behaviors across sequences of actions and manage consumption costs.

### Amazon — Build visibility for Codex on Amazon Bedrock with OpenTelemetry and Amazon CloudWatch

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T16:30:47+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/build-visibility-for-codex-on-amazon-bedrock-with-opentelemetry-and-amazon-cloudwatch/
- Record: https://forck.live/items/3630-build-visibility-for-codex-on-amazon-bedrock-with-opentelemetry-and-amazon
- Subject: Bedrock / Nova

AWS announced a reference pattern to collect OpenTelemetry metrics from Codex clients using Amazon Bedrock and send them to Amazon CloudWatch, enabling visibility into adoption, consumption, and reliability without adding a proxy to the inference path.

### Amazon — Enforcing data residency with single-Region Claude Code on Amazon Bedrock

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T16:21:54+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/enforcing-data-residency-with-single-region-claude-code-on-amazon-bedrock/
- Record: https://forck.live/items/3631-enforcing-data-residency-with-single-region-claude-code-on-amazon-bedrock
- Subject: Bedrock / Nova

Amazon Bedrock introduces two paths to enforce single-Region data residency for Claude Code: Path 1 uses the Mantle endpoint with native single-Region routing in seven Regions (Ireland, Stockholm, Tokyo, Melbourne, US East (N. Virginia), US East (Ohio), US West (Oregon)), and Path 2 uses the classic Invoke API with an application inference profile, currently only supported in London (eu-west-2) for Claude Opus 4.6 and Sonnet 4.6.

### Amazon — Agent Skills for Automated Reasoning policies in Amazon Bedrock

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T16:12:24+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/agent-skills-for-automated-reasoning-policies-in-amazon-bedrock/
- Record: https://forck.live/items/3632-agent-skills-for-automated-reasoning-policies-in-amazon-bedrock
- Subject: Bedrock / Nova

Amazon announces Agent Skills for Automated Reasoning policies in Amazon Bedrock, a suite of six open-format skills that guide coding agents through the full lifecycle of building, testing, deploying, and validating Automated Reasoning policies.

### Amazon — Building an agentic app deployer with Amazon Bedrock and AWS Lambda

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T16:11:13+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/building-an-agentic-app-deployer-with-amazon-bedrock-and-aws-lambda/
- Record: https://forck.live/items/3633-building-an-agentic-app-deployer-with-amazon-bedrock-and-aws-lambda
- Subject: Bedrock / Nova

Amazon Bedrock and AWS Lambda are used to build an agentic app deployer called PDI Brew, which allows non-technical employees to describe internal tools in plain English and have them automatically provisioned as multi-tenant web applications on AWS.

### Amazon — LLM optimization integration for Amazon SageMaker Python SDK

- Company: Amazon (amazon.com)
- Announced: 2026-08-06T16:08:12+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/llm-optimization-integration-for-amazon-sagemaker-python-sdk/
- Record: https://forck.live/items/3634-llm-optimization-integration-for-amazon-sagemaker-python-sdk
- Subject: Bedrock / Nova

Amazon SageMaker Python SDK v3 now includes generative AI inference recommendations directly in notebook workflows, enabling benchmarking, deployment recommendations, and automated deployment of optimal configurations.

### Databricks — Introducing OfficeQA Pro V2: A New Benchmark for Enterprise Grounded-Reasoning

- Company: Databricks (databricks.com)
- Announced: 2026-08-06T16:00:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.databricks.com/blog/introducing-officeqa-pro-v2-new-benchmark-enterprise-grounded-reasoning
- Record: https://forck.live/items/17501-introducing-officeqa-pro-v2-a-new-benchmark-for-enterprise-grounded-reasoning
- Subject: Mosaic AI

Databricks released OfficeQA Pro V2, a benchmark designed to evaluate whether AI agents can generalize to unfamiliar enterprise grounded-reasoning tasks. The benchmark contains 90 questions grounded in approximately 120,000 pages from U.S. Treasury accounts spanning 1793 through 2024. Out-of-the-box frontier agents achieved an average accuracy of 37.5%, while agents developed for the Grounded Reasoning Cup averaged 41.1% and the winning team reached 63.3%.

### Google DeepMind — WeatherNext: AI model achieves breakthrough in forecasting cyclones

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-08-06T15:06:15+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://deepmind.google/blog/weathernext-ai-model-achieves-breakthrough-in-forecasting-cyclones/
- Record: https://forck.live/items/3625-weathernext-ai-model-achieves-breakthrough-in-forecasting-cyclones
- Subject: Gemini

Introduces WeatherNext, an AI model for forecasting cyclones.

### Microsoft — Microsoft named a Leader in the 2026 Gartner® Magic Quadrant™ for AI-Augmented Code Modernization Tools

- Company: Microsoft (microsoft.com)
- Announced: 2026-08-06T15:00:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://azure.microsoft.com/en-us/blog/microsoft-named-a-leader-in-the-2026-gartner-magic-quadrant-for-ai-augmented-code-modernization-tools/
- Record: https://forck.live/items/3652-microsoft-named-a-leader-in-the-2026-gartner-magic-quadrant-for-ai-augmented
- Subject: Azure AI

Microsoft has been named a Leader in the 2026 Gartner Magic Quadrant for AI-Augmented Code Modernization Tools, highlighting its Azure and GitHub Copilot capabilities for modernizing legacy applications.

### Google DeepMind — Our WeatherNext 2 AI model demonstrated a massive leap forward in predicting cyclones.

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-08-06T14:00:00+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://blog.google/innovation-and-ai/models-and-research/google-deepmind/weathernext-2-cyclones/
- Record: https://forck.live/items/3626-our-weathernext-2-ai-model-demonstrated-a-massive-leap-forward-in-predicting
- Subject: Gemini

Google DeepMind's WeatherNext 2 AI model achieved state-of-the-art accuracy in predicting cyclone track, intensity, and wind structure, representing a decade of meteorological progress. The model is now open-sourced to the global research community.

### NVIDIA — Into the Omniverse: How Open World Models Push the Frontier of Physical AI

- Company: NVIDIA (nvidia.com)
- Announced: 2026-08-06T13:00:00+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://blogs.nvidia.com/blog/open-world-models-physical-ai/
- Record: https://forck.live/items/3609-into-the-omniverse-how-open-world-models-push-the-frontier-of-physical-ai
- Subject: AI platform

NVIDIA announces Cosmos 3, an open physical AI foundation model family that combines vision reasoning, world generation, and action prediction. The family includes three sizes: Super (64B), Nano (16B), and Edge (4B), with the latter designed for on-device deployment on edge GPUs.

### OpenAI — Improving GPT‑5.6 Sol in ChatGPT—and expanding access to GPT-5.6 Luna for free users

- Company: OpenAI (openai.com)
- Announced: 2026-08-06T10:00:00+00:00
- Category: model-update
- Coverage: 28 outlets
- Announcement: yes
- Group: models
- Source: https://openai.com/index/improving-gpt-5-6-sol-in-chatgpt
- Record: https://forck.live/items/3622-improving-gpt-5-6-sol-in-chatgpt-and-expanding-access-to-gpt-5-6-luna-for-free
- Subject: GPT / ChatGPT / API

OpenAI announces an improved version of GPT-5.6 Sol in ChatGPT with better accuracy and consistency, and expands access to GPT-5.6 Luna for free users, including unlimited everyday chats.

### OpenAI — Working with the American Psychological Association on youth mental health and AI

- Company: OpenAI (openai.com)
- Announced: 2026-08-06T06:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/openai-and-apa-partner-to-advance-responsible-ai
- Record: https://forck.live/items/3623-working-with-the-american-psychological-association-on-youth-mental-health-and
- Subject: GPT / ChatGPT / API

OpenAI and the American Psychological Association are collaborating to create evidence-based guidance and safeguards for responsible AI use, focusing on youth mental health.

### Vercel — Introducing Agent Plugins

- Company: Vercel (vercel.com)
- Announced: 2026-08-06
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/introducing-agent-plugins
- Record: https://forck.live/items/20677-introducing-agent-plugins
- Subject: v0 / AI SDK / AI Gateway

Vercel announced Agent Plugins 1.0.0, an open, vendor-neutral standard for packaging Agent Skills and MCP servers into distributable plugins. The format defines a directory structure with a plugin.json manifest and fixed locations for components, leaving client-specific behavior to each client. The specification was refined collaboratively with representatives from AWS, Anysphere, GitHub, Microsoft, OpenAI, and Vercel, and is supported at launch by ChatGPT, Codex, Cursor, GitHub Copilot, Kiro, and VS Code.

### Ai2 — Ai2 expands collaboration with Hugging Face to accelerate open science

- Company: Ai2 (allenai.org)
- Announced: 2026-08-06
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://allenai.org/blog/hugging-face-partnership
- Record: https://forck.live/items/18541-ai2-expands-collaboration-with-hugging-face-to-accelerate-open-science
- Subject: Olmo / Molmo / Asta

Ai2 and Hugging Face are expanding their partnership to make fully open AI easier to access and deploy. Hugging Face is roughly tripling Ai2's storage on the Hub to nearly two petabytes and removing standard rate limits on traffic. The collaboration also includes integration of Ai2's open robotics model MolmoAct 2 into Hugging Face's LeRobot library and the adoption of Ai2's olmOCR-Bench as Hugging Face's go-to OCR benchmark.

### Genspark — The AI Efficiency Frontier: How Enterprises Scale Agentic AI Without Letting Cost Outrun Value

- Company: Genspark (genspark.ai)
- Announced: 2026-08-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.genspark.ai/blog/the-ai-efficiency-frontier
- Record: https://forck.live/items/18245-the-ai-efficiency-frontier-how-enterprises-scale-agentic-ai-without-letting
- Subject: Genspark

insights Kay, Co-founder and CTO of Genspark

### Genspark — How This 26-Year-Old Entrepreneur Used Genspark to Build an AI Chief of Staff With No Tech Background

- Company: Genspark (genspark.ai)
- Announced: 2026-08-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.genspark.ai/blog/entrepreneur-builds-ai-chief-of-staff-no-tech-background
- Record: https://forck.live/items/18244-how-this-26-year-old-entrepreneur-used-genspark-to-build-an-ai-chief-of-staff
- Subject: Genspark

### Cua — Blending the best of computer and browser use, without a Chrome extension

- Company: Cua (cua.ai)
- Announced: 2026-08-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cua.ai/blog/extension-free-browser-use
- Record: https://forck.live/items/16817-blending-the-best-of-computer-and-browser-use-without-a-chrome-extension
- Subject: Cua Driver / Lume / Cua Bench / Cua Fleets

Cua launched Cua Driver 0.19.0, an extension-free browser-use interface built into its agent-neutral computer-use driver. The update connects an exact Chromium tab to its native process and window, giving agents page-aware browser actions and full desktop control in the same session.

### Suno — How We're Building the Future of Music Responsibly

- Company: Suno (suno.com)
- Announced: 2026-08-06
- Category: safety-policy-update
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://suno.com/blog/building-the-future-of-music-responsibly
- Record: https://forck.live/items/10315-how-we-re-building-the-future-of-music-responsibly
- Subject: Music / Studio

Suno published its principles for responsible music AI, including a commitment to prevent imitation of existing artists, a forthcoming downloads policy to limit mass distribution on streaming platforms, and new audio watermarking and fingerprinting tools to identify AI-generated songs. The company also updated its Community Guidelines to clarify prohibited activities such as recreating existing songs or using someone's voice without permission.

### Luma — MiniMax H3 Available Now

- Company: Luma (lumalabs.ai)
- Announced: 2026-08-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lumalabs.ai/news/minimax-h3-now-available-in-luma-agents
- Record: https://forck.live/items/7956-minimax-h3-available-now
- Subject: Ray / Dream Machine

Luma released MiniMax H3, a model that generates video and matched audio in a single pass, available to self-serve accounts.

### Luma — Seedance 2.5: Now Live in Luma

- Company: Luma (lumalabs.ai)
- Announced: 2026-08-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lumalabs.ai/news/seedance-2.5-live-in-luma
- Record: https://forck.live/items/7955-seedance-2-5-now-live-in-luma
- Subject: Ray / Dream Machine

Luma added Seedance 2.5, which supports native 30-second video generation in a single pass, up to 50 reference inputs, native audio, style swap, and targeted fixes with improved lighting physics.

### Amp — Portals into Orbs

- Company: Amp (ampcode.com)
- Announced: 2026-08-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/portals
- Record: https://forck.live/items/7718-portals-into-orbs
- Subject: Amp

Amp is shipping portals, a feature that lets users access anything running in an orb that listens on a port and speaks HTTP, with live reloading and no VPN.

### Together AI — DeepSeek-V4 Flash 0731 vs GPT-5.6 Luna on DeepSWE: Cost and Coding

- Company: Together AI (together.ai)
- Announced: 2026-08-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.together.ai/blog/deepseek-v4-flash-0731-vs-gpt-5-6-luna-on-deepswe-cost-and-coding
- Record: https://forck.live/items/3651-deepseek-v4-flash-0731-vs-gpt-5-6-luna-on-deepswe-cost-and-coding
- Subject: Inference platform

The post compares DeepSeek-V4 Flash 0731 and GPT-5.6 Luna on DeepSWE, showing that while GPT-5.6 Luna is more accurate, DeepSeek-V4 Flash is much cheaper, and a cascade strategy combining both outperforms Luna alone on accuracy and cost.

### OpenAI — From asking to doing: How the world is putting ChatGPT to work

- Company: OpenAI (openai.com)
- Announced: 2026-08-06
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/how-the-world-is-putting-chatgpt-to-work
- Record: https://forck.live/items/3624-from-asking-to-doing-how-the-world-is-putting-chatgpt-to-work
- Subject: GPT / ChatGPT / API

OpenAI released Signals data showing global ChatGPT usage patterns with country-level adoption and trend insights.

### Cohere — Cohere and the University of Waterloo launch partnership to strengthen Canada’s AI talent pipeline

- Company: Cohere (cohere.com)
- Announced: 2026-08-06
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cohere.com/blog/cohere-university-of-waterloo-announcement
- Record: https://forck.live/items/3611-cohere-and-the-university-of-waterloo-launch-partnership-to-strengthen-canada
- Subject: Command

Cohere and the University of Waterloo announced a partnership to develop a new AI Transformation and Change Management certificate program, launching in Fall 2026, to give students hands-on experience in AI adoption and organizational change.

### Hugging Face — Baseten on Hugging Face Inference Providers 🔥

- Company: Hugging Face (huggingface.co)
- Announced: 2026-08-06
- Category: infrastructure-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/baseten
- Record: https://forck.live/items/3610-baseten-on-hugging-face-inference-providers
- Subject: Platform

Baseten is now a supported Inference Provider on the Hugging Face Hub, offering serverless inference for conversational and text-generation tasks with models like Kimi K3, DeepSeek V4 Flash, and GLM-5.2.

### Meta — From User Sequences to Scaling Laws: A Multi-Stage Architecture for Meta’s Ads Ranking

- Company: Meta (meta.com)
- Announced: 2026-08-05T19:20:20+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://engineering.fb.com/2026/08/05/ml-applications/from-user-sequences-to-scaling-laws-a-multi-stage-architecture-for-metas-ads-ranking/
- Record: https://forck.live/items/3574-from-user-sequences-to-scaling-laws-a-multi-stage-architecture-for-meta-s-ads
- Subject: Llama / infrastructure

Meta introduces a multi-stage sequence model and dense tokenization with target-aware attention to scale sequence learning for ads ranking, achieving lifts in conversions and clicks.

### Amazon — How LendingTree built a multi-agent mortgage assistant on Amazon Bedrock

- Company: Amazon (amazon.com)
- Announced: 2026-08-05T18:50:02+00:00
- Category: model-update
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/how-lendingtree-built-a-multi-agent-mortgage-assistant-on-amazon-bedrock/
- Record: https://forck.live/items/3575-how-lendingtree-built-a-multi-agent-mortgage-assistant-on-amazon-bedrock
- Subject: Bedrock / Nova

LendingTree built a multi-agent mortgage assistant on Amazon Bedrock, using three agents (supervisor, education, matching) coordinated via LangGraph and MCP. The supervisor agent uses Amazon Nova Pro for complex reasoning and Amazon Nova Lite for simpler tasks, and the system leverages Amazon Bedrock Guardrails for compliance and PII protection.

### Amazon — How Mobileye transformed support operations using Amazon Bedrock AgentCore

- Company: Amazon (amazon.com)
- Announced: 2026-08-05T18:09:03+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/how-mobileye-transformed-support-operations-using-amazon-bedrock-agentcore/
- Record: https://forck.live/items/3571-how-mobileye-transformed-support-operations-using-amazon-bedrock-agentcore
- Subject: Bedrock / Nova

Mobileye used Amazon Bedrock AgentCore to deploy an AI support agent that reduced response times by 90% and achieved over 95% accuracy, with zero infrastructure management, and later turned it into a self-service platform.

### Amazon — How we built an MCP bridge to give our AgentCore-hosted AI agent access to local MCP tools

- Company: Amazon (amazon.com)
- Announced: 2026-08-05T18:02:23+00:00
- Category: infrastructure-release
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/how-we-built-an-mcp-bridge-to-give-our-agentcore-hosted-ai-agent-access-to-local-mcp-tools/
- Record: https://forck.live/items/3572-how-we-built-an-mcp-bridge-to-give-our-agentcore-hosted-ai-agent-access-to
- Subject: Bedrock / Nova

Amazon describes how they built an MCP bridge to enable their AgentCore-hosted AI agent to access local MCP tools, using WebSocket and native messaging to tunnel MCP messages between the remote client and the local server.

### Amazon — Run production AI agents in n8n with Amazon Bedrock AgentCore harness

- Company: Amazon (amazon.com)
- Announced: 2026-08-05T18:00:57+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://aws.amazon.com/blogs/machine-learning/run-production-ai-agents-in-n8n-with-amazon-bedrock-agentcore-harness/
- Record: https://forck.live/items/3573-run-production-ai-agents-in-n8n-with-amazon-bedrock-agentcore-harness
- Subject: Bedrock / Nova

Amazon Bedrock AgentCore's harness is now generally available, with a new open-source community node for n8n that allows building production AI agents with persistent memory, tools, and model choice without writing infrastructure code.

### Harvey — The Ultimate Guide to Landing a Job at Harvey

- Company: Harvey (harvey.ai)
- Announced: 2026-08-05T16:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/landing-a-job-at-harvey
- Record: https://forck.live/items/17536-the-ultimate-guide-to-landing-a-job-at-harvey
- Subject: Harvey

Our Chief Operating Officer answers the top 25 questions she gets from candidates. At Harvey, we’re very lucky to be hiring across most of our teams.

### Tencent — From LR to ELR: A Better Heuristic for Pretraining Dynamics

- Company: Tencent (tencent.com)
- Announced: 2026-08-05T16:00:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://hunyuan.tencent.com/research/elr
- Record: https://forck.live/items/4576-from-lr-to-elr-a-better-heuristic-for-pretraining-dynamics
- Subject: Hunyuan

Tencent Hunyuan research introduces the Effective Learning Rate (ELR) as a better heuristic for controlling angular update size (AUS) during LLM pretraining. The post shows that ELR leads to Hyperball optimization, enables more accurate loss prediction, and the proposed minus-square-root schedule sets a new SOTA on the Modded-nanoGPT Track 3 benchmark, reaching target loss 3.28 in 3175 steps.

### Krea — Train in cluster, infer anywhere

- Company: Krea (krea.ai)
- Announced: 2026-08-05T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.krea.ai/blog/train-in-cluster-infer-anywhere
- Record: https://forck.live/items/15100-train-in-cluster-infer-anywhere
- Subject: Krea 2 / AI creative platform

Krea describes how it built a system so inference keeps running on external infrastructure when training jobs claim all local GPUs. The solution uses the Virtual Kubelet project to emulate a Kubernetes node with effectively infinite GPU capacity, abstracting business logic into External Providers that handle pod lifecycle on various real backends.

### Lovable — Lovable + Cerebras: investing in infrastructure that keeps up with your ideas

- Company: Lovable (lovable.dev)
- Announced: 2026-08-05T13:25:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/cerebras-partnership
- Record: https://forck.live/items/12063-lovable-cerebras-investing-in-infrastructure-that-keeps-up-with-your-ideas
- Subject: Lovable / AI app builder

Lovable is partnering with Cerebras to reduce response times by 2027, using Cerebras' Wafer-Scale Engine to keep models on a single chip and cut the 'communication tax' of traditional GPUs. The company will move onto dedicated Cerebras capacity in phases, starting with its most latency-sensitive workloads.

### Sarvam AI — Introducing Sarvam Circle

- Company: Sarvam AI (sarvam.ai)
- Announced: 2026-08-05T12:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.sarvam.ai/blogs/introducing-sarvam-circle/
- Record: https://forck.live/items/18386-introducing-sarvam-circle
- Subject: Sarvam / Bulbul / Saaras

Sarvam AI announced Sarvam Circle, a partnership program for cloud providers, OEMs, system integrators, consulting firms, ISVs, and investors. The program aims to co-create industry solutions on Sarvam's AI stack and scale pilots into production across Indian institutions. Partners include AWS, HCLTech, Thoughtworks, IBM, HP, and YCP India.

### Lovable — A dedicated security page for every business built on Lovable

- Company: Lovable (lovable.dev)
- Announced: 2026-08-05T12:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/trust-centers-for-business
- Record: https://forck.live/items/12064-a-dedicated-security-page-for-every-business-built-on-lovable
- Subject: Lovable / AI app builder

Lovable launched trust centers for every app published on its platform. Each app gets a dedicated security page at its own URL that automatically displays active security controls such as HTTPS enforcement, TLS certificate status, and dependency vulnerability checks, without requiring manual input from the developer.

### Vercel — Introducing the new v0 API

- Company: Vercel (vercel.com)
- Announced: 2026-08-05T04:00:00+00:00
- Category: api-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/introducing-the-new-v0-api
- Record: https://forck.live/items/20678-introducing-the-new-v0-api
- Subject: v0 / AI SDK / AI Gateway

Vercel released a new API for v0 that provides programmatic, headless access to its app-building agent. The API allows developers to send a prompt, receive a generated app with a dev server in a Vercel Sandbox, and embed the preview URL in their own UI. It supports synchronous, asynchronous, and streaming responses, and is now generally available.

### Tencent — Tencent Hy3 Now Available Globally, Extending Practical AI Across Products, Workflows and Cloud Services

- Company: Tencent (tencent.com)
- Announced: 2026-08-05T02:48:04+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.tencent.com/tencent-hy3-now-available-globally-extending-practical-ai-across-products-workflows-and-cloud-services/
- Record: https://forck.live/items/4569-tencent-hy3-now-available-globally-extending-practical-ai-across-products
- Subject: Hunyuan

Tencent announced broader international access to the Hy3 model, making it available globally through WorkBuddy, Miora, and Tencent Cloud TokenHub, with free access on WorkBuddy until 31 August 2026.

### Supabase — Postgres Changes gets AND filters, new operators, and column selection

- Company: Supabase (supabase.com)
- Announced: 2026-08-05
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/postgres-changes-filters-and-column-selection
- Record: https://forck.live/items/21274-postgres-changes-gets-and-filters-new-operators-and-column-selection
- Subject: Supabase

Supabase added three capabilities to Postgres Changes: filters that combine across multiple columns, additional filter operators (like, ilike, is, match, imatch, isdistinct, each invertible with not), and column selection so a subscription returns only the requested fields. All three are available now.

### Liquid AI — MacPaw Partners with Liquid AI to Bring On-Device AI to Millions of Mac Users

- Company: Liquid AI (liquid.ai)
- Announced: 2026-08-05
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.liquid.ai/blog/macpaw-partners-liquid-ai-on-device-ai-mac-users
- Record: https://forck.live/items/16775-macpaw-partners-with-liquid-ai-to-bring-on-device-ai-to-millions-of-mac-users
- Title in English (translated by forck.live from en): MacPaw Partners with Liquid AI to Bring On-Device AI to Millions of Mac Users
- Subject: LFM / d1 models

MacPaw and Liquid AI announced a strategic partnership to co-develop a local AI stack for the Mac, combining Liquid Foundation Models with MacPaw's on-device inference (Elix) and memory (Mnemos) technologies. The collaboration will first bring on-device intelligence, persistent memory, and native task execution to Eney, MacPaw's AI assistant for macOS, with results expected later this year. The technologies are designed as shared infrastructure that could eventually be made available to Mac developers through Setapp.

### Suno — How to make short-form content for your songs

- Company: Suno (suno.com)
- Announced: 2026-08-05
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://suno.com/blog/how-to-make-short-form
- Record: https://forck.live/items/10316-how-to-make-short-form-content-for-your-songs
- Subject: Music / Studio

Suno's guide offers practical advice for musicians on creating short-form video content to promote their songs, covering hooks, visual energy matching, formats, storytelling, phased promotion, series ideas, and distribution basics.

### Meta — Introducing Muse Code and Muse Spark 1.2

- Company: Meta (meta.com)
- Announced: 2026-08-05
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://research.meta.ai/blog/introducing-muse-code-and-muse-spark-1-2
- Record: https://forck.live/items/7280-introducing-muse-code-and-muse-spark-1-2
- Subject: Muse

Meta announces Muse Code, a terminal coding agent powered by the new Muse Spark 1.2 model, with improvements in code generation and debugging.

### LG AI Research — [CVPR 2026] Beyond Video World Models: What Should We Predict About the Future?

- Company: LG AI Research (lgresearch.ai)
- Announced: 2026-08-05
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.lgresearch.ai/blog/view?seq=671
- Record: https://forck.live/items/4738-cvpr-2026-beyond-video-world-models-what-should-we-predict-about-the-future
- Subject: EXAONE

LG AI Research discusses world model research at CVPR 2026, examining different approaches to predicting the future in AI, from video generation to latent representations, and introduces their own work 'Align While Search' (AWS) for decision-making under partial observability.

### AMD — AMD Reports Second Quarter 2026 Financial Results

- Company: AMD (amd.com)
- Announced: 2026-08-04T20:15:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ir.amd.com/news-events/press-releases/detail/1295/amd-reports-second-quarter-2026-financial-results
- Record: https://forck.live/items/15034-amd-reports-second-quarter-2026-financial-results
- Subject: AMD Instinct / ROCm

AMD reported second quarter 2026 revenue of $11.5 billion, up 50% year-over-year, with Data Center segment revenue of $6.7 billion more than doubling from the prior year. GAAP net income was $2.3 billion and diluted earnings per share was $1.38.

### GitHub — Retiring the Copilot Billing Preview app

- Company: GitHub (github.com)
- Announced: 2026-08-04T19:09:32+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-04-retiring-the-copilot-billing-preview-app
- Record: https://forck.live/items/3558-retiring-the-copilot-billing-preview-app
- Subject: Copilot

GitHub has retired the Copilot Billing Preview app. Users should now use the billing settings directly to review and manage Copilot spend.

### OpenAI — Third-party cyber evaluations involving OpenAI models

- Company: OpenAI (openai.com)
- Announced: 2026-08-04T19:00:00+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/third-party-cyber-evaluations-involving-openai-models
- Record: https://forck.live/items/3566-third-party-cyber-evaluations-involving-openai-models
- Subject: GPT / ChatGPT / API

OpenAI explains recent third-party cybersecurity evaluation incidents and outlines new safeguards to strengthen AI model testing and evaluation.

### Amazon — Introducing Web Search on Amazon Bedrock for foundation model grounding

- Company: Amazon (amazon.com)
- Announced: 2026-08-04T18:39:14+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/introducing-web-search-on-amazon-bedrock-for-foundation-model-grounding/
- Record: https://forck.live/items/3554-introducing-web-search-on-amazon-bedrock-for-foundation-model-grounding
- Subject: Bedrock / Nova

Amazon announced general availability of Web Search on Amazon Bedrock, a server-side built-in tool that grounds model responses in current web knowledge using Amazon's own web index and knowledge graph, with no third-party vendors or external APIs required. The tool supports multi-source grounding, context-efficient retrieval via semantic snippet extraction, single-parameter enablement in OpenAI-compatible API calls, and zero data egress for enterprise compliance.

### Amazon — Automated web insight extraction with Amazon Bedrock AgentCore

- Company: Amazon (amazon.com)
- Announced: 2026-08-04T16:02:21+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/automated-web-insight-extraction-with-amazon-bedrock-agentcore/
- Record: https://forck.live/items/3547-automated-web-insight-extraction-with-amazon-bedrock-agentcore
- Subject: Bedrock / Nova

Amazon Bedrock AgentCore introduces a fully managed browser service that renders JavaScript-heavy pages reliably, enabling automated web insight extraction pipelines.

### NVIDIA — NVIDIA Joins NSF State and Regional AI Hubs Program to Expand AI Research and Education Across the US

- Company: NVIDIA (nvidia.com)
- Announced: 2026-08-04T16:00:56+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://blogs.nvidia.com/blog/nsf-state-regional-ai-hub-program/
- Record: https://forck.live/items/3546-nvidia-joins-nsf-state-and-regional-ai-hubs-program-to-expand-ai-research-and
- Subject: AI platform

NVIDIA is participating in the NSF State and Regional AI Hubs program to expand access to AI computing, data, software, and expertise for research and education across the US.

### ByteDance — SeedRealtime Audio-Visual Full-Duplex LLM Released: Toward Omni-Modal Natural Interaction

- Company: ByteDance (bytedance.com)
- Announced: 2026-08-04T16:00:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://seed.bytedance.com/en/blog/seedrealtime-audio-visual-full-duplex-llm-released-toward-omni-modal-natural-interaction
- Record: https://forck.live/items/4658-seedrealtime-audio-visual-full-duplex-llm-released-toward-omni-modal-natural
- Subject: Seed / Doubao

ByteDance has released SeedRealtime, a native audio-visual full-duplex LLM designed for omni-modal natural interaction, featuring joint audio-visual understanding, proactive interaction, and natural conversational timing. The model has been fully rolled out with large-scale deployment.

### GitHub — Upcoming deprecation of GitHub Spark on github.com

- Company: GitHub (github.com)
- Announced: 2026-08-04T15:54:47+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://github.blog/changelog/2026-08-04-upcoming-deprecation-of-github-spark-on-github-com
- Record: https://forck.live/items/3548-upcoming-deprecation-of-github-spark-on-github-com
- Subject: Copilot

Beginning August 4, 2026, GitHub Spark no longer accepts new users or allows the creation of new apps. Existing users can continue to access GitHub Spark until August 31, 2026 to export apps they’ve already created. Apps you’ve already deployed will continue to work after GitHub Spark is retired. GitHub Spark was created to help people quickly move from an idea to a working application. Since its launch, AI models and agentic development tools have advanced significantly. Builders can now build and refine these experiences through GitHub Copilot in the environments where they already work, including VS Code, Copilot CLI, and the GitHub Copilot app. We’ve seen builders increasingly choose these integrated workflows for application development, and we’re aligning our experiences accordingly. This change applies specifically to the current GitHub Spark experience on github.com. Separately, GitHub Models, the inference service used by Spark’s llm() function, retired on July 30, 2026 . As of that date, calls to llm() no longer work. Apps that don’t use llm() aren’t affected by the GitHub Models retirement. What this means for you …

### NVIDIA — NVIDIA Alpamayo 2 Super, the Frontier Open Model for Robotaxis and Autonomous Vehicles, Now Available for Commercial Use

- Company: NVIDIA (nvidia.com)
- Announced: 2026-08-04T15:00:49+00:00
- Category: not stated
- Coverage: 3 outlets
- Announcement: no
- Group: covered
- Source: https://blogs.nvidia.com/blog/alpamayo-2-super-open-model-now-available/
- Record: https://forck.live/items/3497-nvidia-alpamayo-2-super-the-frontier-open-model-for-robotaxis-and-autonomous
- Subject: AI platform

For robotaxis and other autonomous vehicles (AVs), the hardest problems aren’t the everyday scenarios. They’re the rare, complex situations that are difficult to anticipate and train for. Handling these long‑tail events takes more than just object detection and motion prediction. AVs must understand the situation, reason about cause and effect, choose the right action and turn that decision into a safe, comfortable path — all in real time and in a way developers can inspect, validate and trust. NVIDIA Alpamayo 2 Super, available now for commercial use, is part of the Alpamayo family, the most-adopted open reasoning models for autonomous driving on Hugging Face, supporting a wide range of AV-relevant capabilities within a single foundation model. Built on NVIDIA Cosmos 3 Super Reasoner and post‑trained with reinforcement learning, the model advances the AV ecosystem on two fronts: open commercial licensing and leading multitask capabilities for autonomous driving. Alpamayo 2 Super is part of NVIDIA’s growing collection of open models, datasets and tools for autonomous driving …

### NVIDIA — As AI Increases Demands on Memory, Storage Steps Up

- Company: NVIDIA (nvidia.com)
- Announced: 2026-08-04T15:00:47+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://blogs.nvidia.com/blog/ai-storage-fms/
- Record: https://forck.live/items/3498-as-ai-increases-demands-on-memory-storage-steps-up
- Subject: AI platform

NVIDIA announced open sourcing of cuFile APIs and the Storage-Next initiative to improve AI storage performance and security.

### Black Forest Labs — FLUX 3 Video, Part 1: Generation

- Company: Black Forest Labs (bfl.ai)
- Announced: 2026-08-04T15:00:00+00:00
- Category: model-update
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://bfl.ai/blog/flux-3-video
- Record: https://forck.live/items/3553-flux-3-video-part-1-generation
- Subject: FLUX

Black Forest Labs announced the general availability of FLUX 3 Video, a multimodal model that generates video clips up to 20 seconds with native audio, supporting text-to-video, image-to-video, and video continuation features.

### NVIDIA — AI Leaders Propose SAFE Guidelines for Cybersecurity Transparency

- Company: NVIDIA (nvidia.com)
- Announced: 2026-08-04T13:00:47+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://blogs.nvidia.com/blog/open-secure-ai-alliance-contributions/
- Record: https://forck.live/items/3499-ai-leaders-propose-safe-guidelines-for-cybersecurity-transparency
- Subject: AI platform

SAFE is a proposed set of guidelines for sharing AI cybersecurity incidents confidentially across the ecosystem, aiming to reduce systemic risk through collective defense.

### Harvey — Turn Your Standards Into Stronger Reviews

- Company: Harvey (harvey.ai)
- Announced: 2026-08-04T12:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/playbook-builder-in-harvey
- Record: https://forck.live/items/17537-turn-your-standards-into-stronger-reviews
- Subject: Harvey

Harvey introduced a playbook builder that creates review-ready playbooks from existing standards, past documents, or through a guided conversation. The tool analyzes sources, asks clarifying questions, and drafts rules in real time, incorporating methodology from over 300 customer playbooks. It includes quality controls such as coverage gap detection and source citations, and allows updates to fallbacks or liability caps without rebuilding rules.

### Anthropic — Responsible Scaling Policy

- Company: Anthropic (anthropic.com)
- Announced: not stated by the source (first seen 2026-08-04T07:42:31.596929+00:00)
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.anthropic.com/news/announcing-our-updated-responsible-scaling-policy
- Record: https://forck.live/items/3262-responsible-scaling-policy
- Subject: Claude

Announcements Announcing our updated Responsible Scaling Policy Oct 15, 2024 Read the Responsible Scaling Policy Today we are publishing a significant update to our Responsible Scaling Policy (RSP), the risk governance framework we use to mitigate potential catastrophic risks from frontier AI systems. This update introduces a more flexible and nuanced approach to assessing and managing AI risks while maintaining our commitment not to train or deploy models unless we have implemented adequate safeguards. Key improvements include new capability thresholds to indicate when we will upgrade our safeguards, refined processes for evaluating model capabilities and the adequacy of our safeguards (inspired by safety case methodologies ), and new measures for internal governance and external input. By learning from our implementation experiences and drawing on risk management practices used in other high-consequence industries, we aim to better prepare for the rapid pace of AI advancement. The promise and challenge of advanced AI As frontier AI models advance, they have the potential to bring about transformative benefits for our society and economy. …

### Vercel — Next.js 16.3 support on Vercel

- Company: Vercel (vercel.com)
- Announced: 2026-08-04T07:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/vercel-supports-next-js-16-3
- Record: https://forck.live/items/20679-next-js-16-3-support-on-vercel
- Subject: v0 / AI SDK / AI Gateway

Vercel announced full support for Next.js 16.3, including immutable static assets, faster routing, and improved observability. Upgraded apps saw 45% fewer prefetch requests, 17% fewer CDN requests, and up to 60% global TTFB reductions. Vercel also launched observability for Partial Prerendering (PPR).

### Gradium — New voices: how we pick the one that wins

- Company: Gradium (gradium.ai)
- Announced: 2026-08-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://gradium.ai/blog/new-voices-how-we-pick-the-one-that-wins
- Record: https://forck.live/items/18308-new-voices-how-we-pick-the-one-that-wins
- Subject: Gradium TTS / Phonon

Gradium describes its process for selecting and ranking synthetic voices. The pipeline generates hundreds of candidates per target category, then narrows them through listener voting, ELO-based head-to-head ranking against existing flagship voices, and native-speaker validation. The post also provides a five-step guide for enterprise customers to define voice targets.

### Factory AI — Factory Joins Open Weights and American AI Leadership

- Company: Factory AI (factory.com)
- Announced: 2026-08-04
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://factory.com/news/open-weights-american-ai-leadership
- Record: https://forck.live/items/17864-factory-joins-open-weights-and-american-ai-leadership
- Subject: Droid

Factory signed the Open Weights and American AI Leadership letter, joining Microsoft, NVIDIA, and 230 other organizations. The company states its commitment to an open ecosystem, model independence, and open weight security models.

### Liquid AI — LFM2.5-2.6B: Deploy Agents Everywhere

- Company: Liquid AI (liquid.ai)
- Announced: 2026-08-04
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.liquid.ai/blog/lfm2-5-2-6b
- Record: https://forck.live/items/16776-lfm2-5-2-6b-deploy-agents-everywhere
- Subject: LFM / d1 models

Liquid AI released LFM2.5-2.6B, a 2.6 billion parameter model designed for on-device agentic workloads such as planning, tool calling, and multi-step tasks. The model is pre-trained on approximately 34 trillion tokens, supports a 128K token context window, and is available on Hugging Face in both base and post-trained versions. Benchmarks show it is competitive with models nearly four times its size on STEM, instruction following, tool use, and agentic tasks.

### Amp — Attach Anything

- Company: Amp (ampcode.com)
- Announced: 2026-08-04
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/attach-anything
- Record: https://forck.live/items/7719-attach-anything
- Subject: Amp

Amp launched a new capability that lets users upload any file type—videos, logs, PDFs, spreadsheets, datasets, and more—to their orbs for the system to see and use, enabling tasks such as debugging from log files, converting presentations into interactive websites, generating videos and CAD files, and transcribing media.

### Replicate — The Ultimate Guide to FLUX 3

- Company: Replicate (replicate.com)
- Announced: 2026-08-04
- Category: new-model
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replicate.com/blog/flux-3
- Record: https://forck.live/items/4387-the-ultimate-guide-to-flux-3
- Subject: Platform

Black Forest Labs released FLUX 3, a new multimodal foundation model capable of text-to-video, image-to-video, and video continuation, with a consolidated architecture trained on images, audio, and video.

### OpenAI — New ways to learn and teach with ChatGPT Work and Codex

- Company: OpenAI (openai.com)
- Announced: 2026-08-04
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/learn-teach-chatgpt-work-codex
- Record: https://forck.live/items/3550-new-ways-to-learn-and-teach-with-chatgpt-work-and-codex
- Subject: GPT / ChatGPT / API

OpenAI announces new education plugins for ChatGPT Work and Codex aimed at K-12 teachers, college educators, and students to support learning, teaching, research, and building.

### Anthropic — Mariano-Florentino (Tino) Cuéllar to join Anthropic as Chief Global Affairs Officer

- Company: Anthropic (anthropic.com)
- Announced: 2026-08-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.anthropic.com/news/tino-cuellar
- Record: https://forck.live/items/3549-mariano-florentino-tino-cu-llar-to-join-anthropic-as-chief-global-affairs
- Subject: Claude

Mariano-Florentino (Tino) Cuéllar will join Anthropic as its first Chief Global Affairs Officer, leading the company’s work on policy, strategic international engagement, and government relationships worldwide. Tino’s career spans law, technology, international security, and public institutions at the international, national, and state levels. He recently stepped down as President of the Carnegie Endowment for International Peace, a leading independent global policy research institution with scholars in 20 countries. Prior to his role at Carnegie, Tino was a Justice of the Supreme Court of California, where his opinions addressed technology and privacy, international agreements, and the separation of powers, among other issues. He was previously director of Stanford's Freeman Spogli Institute for International Studies, co-director of the university’s Center for International Security and Cooperation, and director of the Stanford Cyber Initiative. …

### Mistral AI — Introducing Shieldstral.

- Company: Mistral AI (mistral.ai)
- Announced: 2026-08-04
- Category: product-launch
- Coverage: 3 outlets
- Announcement: yes
- Group: covered
- Source: https://mistral.ai/news/shieldstral
- Record: https://forck.live/items/3504-introducing-shieldstral
- Subject: Mistral / Le Chat

By Mistral Back to Blog 5 min read Share this post Copy url to clipboard Copied Thinking Summary Shieldstral introduces a 3B open-weights multimodal safety classifier that outperforms models up to 7x its size by framing content moderation as a policy-adaptive question-answering task. Unlike traditional guardrail models, it accepts plain-language policies at inference time, unifying text and image safety evaluation without retraining. Released under Apache 2.0, it delivers calibrated safety scores across diverse benchmarks while running efficiently on a single 16GB NVIDIA GPU. A 3B open-weights, policy-adaptive multimodal safety classifier that matches models up to 7x its size on text safety and sets a new state of the art on multimodal moderation. “Does this content promote violence against a protected group? Is this image safe to show to a minor? Did the assistant refuse the request?” Every product that ships a model needs to answer questions like these — but the right answer depends on the product, the audience, and the moment. …

### GitHub — Customize the reasoning level for Copilot cloud agent

- Company: GitHub (github.com)
- Announced: 2026-08-03T23:27:35+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://github.blog/changelog/2026-08-03-customize-the-reasoning-level-for-copilot-cloud-agent
- Record: https://forck.live/items/3492-customize-the-reasoning-level-for-copilot-cloud-agent
- Subject: Copilot

GitHub Copilot cloud agent now allows users to set the reasoning level for models that support it, controlling how much the model reasons before responding.

### GitHub — Enterprise team specialization for managed settings

- Company: GitHub (github.com)
- Announced: 2026-08-03T22:55:29+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://github.blog/changelog/2026-08-03-enterprise-team-specialization-for-managed-settings
- Record: https://forck.live/items/3493-enterprise-team-specialization-for-managed-settings
- Subject: Copilot

GitHub introduces enterprise team specialization for managed settings, allowing administrators to customize Copilot configurations per team using itemized files.

### OpenAI — Apple is getting this wrong

- Company: OpenAI (openai.com)
- Announced: 2026-08-03T22:00:00+00:00
- Category: not stated
- Coverage: 30 outlets
- Announcement: no
- Group: covered
- Source: https://openai.com/index/apple-is-getting-this-wrong
- Record: https://forck.live/items/3430-apple-is-getting-this-wrong
- Subject: GPT / ChatGPT / API

OpenAI responds to Apple's lawsuit, correcting claims about its employees and sharing messages to document the events.

### GitHub — Trigger Copilot automations with comments

- Company: GitHub (github.com)
- Announced: 2026-08-03T19:14:58+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://github.blog/changelog/2026-08-03-trigger-copilot-automations-with-comments
- Record: https://forck.live/items/3494-trigger-copilot-automations-with-comments
- Subject: Copilot

GitHub announces Copilot cloud agent automations triggered by issue or PR comments.

### Runway — How EA Is Bringing Generative AI Into Living, Playable Worlds

- Company: Runway (runway.com)
- Announced: 2026-08-03T19:08:28.341000+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://runway.com/news/company-news/electronic-arts-ai-summit-2026
- Record: https://forck.live/items/7859-how-ea-is-bringing-generative-ai-into-living-playable-worlds
- Subject: Gen / Aleph

Electronic Arts (EA) Chief Strategy Officer Mihir Vaidya discusses how video games historically trained AI and argues that the next frontier involves neuro-symbolic architectures that combine generative AI with determinism and control, rather than relying solely on neural networks. He emphasizes that games require real-time, synchronized, interactive responses—not just realistic visuals—and that control over outputs is a key challenge. EA categorizes AI's impact on games into three horizons: efficiency, expansion, and transformation into new entertainment forms.

### Meta — GEM Training: How Meta Doubled the Efficiency of Its LLM-Scale Ads Foundation Model

- Company: Meta (meta.com)
- Announced: 2026-08-03T18:00:17+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://engineering.fb.com/2026/08/03/ml-applications/training-gem-at-llm-scale-meta-ads-recommendation-foundation-model/
- Record: https://forck.live/items/1420-gem-training-how-meta-doubled-the-efficiency-of-its-llm-scale-ads-foundation
- Subject: Llama / infrastructure

Meta has improved the training efficiency of its Generative Ads Recommendation Model (GEM) by doubling end-to-end training efficiency to 20–25% Model FLOPs Utilization (MFU) while scaling training FLOPs 4x over 12 months.

### Amazon — From weeks to minutes: How Formula 1® uses agentic AI on AWS to accelerate data operations

- Company: Amazon (amazon.com)
- Announced: 2026-08-03T17:24:15+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://aws.amazon.com/blogs/machine-learning/from-weeks-to-minutes-how-formula-1-uses-agentic-ai-on-aws-to-accelerate-data-operations/
- Record: https://forck.live/items/2425-from-weeks-to-minutes-how-formula-1-uses-agentic-ai-on-aws-to-accelerate-data
- Subject: Bedrock / Nova

AWS and Formula 1 launched the Data Accelerator, an agentic AI solution on Amazon Bedrock AgentCore that automates data source onboarding and anomaly detection.

### Runway — How Legal Leaders at Block and Spotify Get to Yes on AI

- Company: Runway (runway.com)
- Announced: 2026-08-03T17:05:03.676000+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://runway.com/news/company-news/spotify-block-ai-summit-2026
- Record: https://forck.live/items/7860-how-legal-leaders-at-block-and-spotify-get-to-yes-on-ai
- Subject: Gen / Aleph

Block and Spotify legal leaders Chrysty Esperanza and Kevan Choset argue that blocking AI is riskier than enabling it, and that legal teams should embed themselves in product development, write internal rules before regulators act, and build tools to streamline AI governance.

### Amazon — Automated Reasoning policy refinement in Amazon Bedrock

- Company: Amazon (amazon.com)
- Announced: 2026-08-03T16:30:58+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://aws.amazon.com/blogs/machine-learning/automated-reasoning-policy-refinement-in-amazon-bedrock/
- Record: https://forck.live/items/2426-automated-reasoning-policy-refinement-in-amazon-bedrock
- Subject: Bedrock / Nova

Amazon Bedrock announces automatic policy refinement for Automated Reasoning checks, with two modes: Iterative Refinement for rule issues and Ambiguous Variable Refinement for language issues.

### Replit — AI adoption starts with truth

- Company: Replit (replit.com)
- Announced: 2026-08-03T16:15:17+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replit.com/blog/ai-adoption
- Record: https://forck.live/items/17891-ai-adoption-starts-with-truth
- Subject: Replit Agent

Replit argues that AI adoption in companies is limited by trust, and that a semantic layer—defining which data tables are sources of truth and how they relate—is necessary for agents to reliably answer business questions. The post describes Replit's approach of building a version-controlled, human-reviewed corpus of definitions and corrections, and notes that Anthropic, OpenAI, and Meta are converging on similar ideas.

### Microsoft Research — Orchard: An open framework for scalable agentic AI

- Company: Microsoft Research (microsoft.com)
- Announced: 2026-08-03T16:00:00+00:00
- Category: open-weight-release
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.microsoft.com/en-us/research/blog/orchard-an-open-framework-for-scalable-agentic-ai/
- Record: https://forck.live/items/2455-orchard-an-open-framework-for-scalable-agentic-ai
- Subject: Research / Phi

Microsoft Research released Orchard, an open-source framework for scalable agentic AI, including three domain-specific training recipes (Orchard-SWE, Orchard-GUI, Orchard-Claw) and associated training data and evaluation methods.

### Runway — Netflix and FutureCel Still Let the Story Be the Why

- Company: Runway (runway.com)
- Announced: 2026-08-03T15:36:30.382000+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://runway.com/news/company-news/netflix-aif-2026
- Record: https://forck.live/items/7861-netflix-and-futurecel-still-let-the-story-be-the-why
- Subject: Gen / Aleph

Netflix's Senior Director of Creative Innovation Girish Balakrishnan and FutureCel founder Joel Kuwahara discussed the role of AI in filmmaking and animation at the 2026 Runway AI Festival, emphasizing that tools should serve storytelling.

### OpenAI — How we built a realtime system for responsive voice AI in six months

- Company: OpenAI (openai.com)
- Announced: 2026-08-03T07:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/continuous-voice-interaction-with-gpt-live
- Record: https://forck.live/items/1-how-we-built-a-realtime-system-for-responsive-voice-ai-in-six-months
- Subject: GPT / ChatGPT / API

OpenAI introduced GPT-Live, a system for continuous voice interaction with AI, using a turnless speech model and low-latency architecture for faster, more natural conversations.

### Vercel — How Factory scaled its cloud backend to one billion monthly requests on Vercel

- Company: Vercel (vercel.com)
- Announced: 2026-08-03T04:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/how-factory-scaled-its-cloud-backend-to-one-billion-monthly-requests-on-vercel
- Record: https://forck.live/items/20680-how-factory-scaled-its-cloud-backend-to-one-billion-monthly-requests-on-vercel
- Subject: v0 / AI SDK / AI Gateway

Factory runs its entire cloud backend on a single Next.js application on Vercel, handling one billion daily requests with a p95 response time of 350ms. Non-technical teams use Factory's own Droids to build and deploy internal tools via Vercel's APIs without engineering involvement. Vercel's Observability and Web Application Firewall provide route-level protection against automated attacks.

### Alibaba — Qwen3.8-Max: A New Bar for Coding and Cowork

- Company: Alibaba (alibaba.com)
- Announced: 2026-08-03T02:00:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://qwen.ai/blog?id=qwen3.8
- Record: https://forck.live/items/4620-qwen3-8-max-a-new-bar-for-coding-and-cowork
- Subject: Qwen

Alibaba releases Qwen 3.8-Max, a 2.4 trillion parameter model (95B active) with improvements across coding, work, research, and long-horizon tasks. Open weights will be released next week. The model demonstrated autonomous coding capabilities, including reproducing research papers and beating human teams in a competition.

### Genspark — GenOffice: The First Open-Source AI Office Suite

- Company: Genspark (genspark.ai)
- Announced: 2026-08-03
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.genspark.ai/blog/genoffice-open-source-ai-office
- Record: https://forck.live/items/18246-genoffice-the-first-open-source-ai-office-suite
- Subject: Genspark

### Factory AI — Designing an AI-Native Enterprise Organizational Model

- Company: Factory AI (factory.com)
- Announced: 2026-08-03
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://factory.com/news/enterprise-organization-model
- Record: https://forck.live/items/17832-designing-an-ai-native-enterprise-organizational-model
- Subject: Droid

Factory describes its Organization Model for deploying AI agents within large, regulated enterprises. The model separates a single enterprise account (billing, identity, administration) from nested organizations that define operating boundaries for teams, regions, or compliance requirements. The Enterprise Admin Console provides centralized visibility and per-organization controls for integrations, model access, usage limits, and data residency.

### Manus — Introducing ElevenLabs on Manus: Give Your Ideas a Voice From One Chat

- Company: Manus (manus.im)
- Announced: 2026-08-03
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://manus.im/blog/elevenlabs-connector
- Record: https://forck.live/items/14702-introducing-elevenlabs-on-manus-give-your-ideas-a-voice-from-one-chat
- Subject: Manus

Manus has introduced a connector for ElevenLabs, allowing users to generate speech, transcribe recordings, clone voices, and build audio applications through a single chat interface. The integration works with an existing ElevenLabs account and does not require coding or API key management. Available capabilities depend on the user's ElevenLabs plan and credit balance.

### OpenAI — Circles powers telco personalization with OpenAI technology

- Company: OpenAI (openai.com)
- Announced: 2026-08-03
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/circles
- Record: https://forck.live/items/3431-circles-powers-telco-personalization-with-openai-technology
- Subject: GPT / ChatGPT / API

Circles leverages OpenAI API and Codex for telco personalization, achieving measurable improvements in ARPU, churn reduction, and dev efficiency.

### Cursor — Google Workspace Plugins

- Company: Cursor (cursor.com)
- Announced: 2026-08-03
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/google-workspace-plugins
- Record: https://forck.live/items/2587-google-workspace-plugins
- Subject: Cursor

Cursor released plugins that give coding agents direct access to Gmail, Google Drive, and Calendar, enabling them to read, write, and act across Google Workspace without leaving the editor.