# forck.live — 1–7 June 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-W23, 1–7 June 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-W23
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Harvey — AI Governance at Harvey: Announcing our ISO 42001 Certification

- Company: Harvey (harvey.ai)
- Announced: 2026-06-05T15:00:00+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/governance-iso-42001
- Record: https://forck.live/items/17554-ai-governance-at-harvey-announcing-our-iso-42001-certification
- Subject: Harvey

Harvey announced it has achieved ISO 42001 certification, an internationally recognized standard for AI management systems. The certification formalizes governance practices including AI risk management, transparency, accountability, and human oversight that Harvey has been building since its inception. The company states the certification provides independent validation of its AI governance practices for enterprise legal customers.

### Google DeepMind — The latest AI news we announced in May 2026

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-06-05T14:45:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://blog.google/innovation-and-ai/technology/ai/google-ai-updates-may-2026/
- Record: https://forck.live/items/1307-the-latest-ai-news-we-announced-in-may-2026
- Subject: Gemini

The source is a title and short description of a blog post listing Google's AI updates from May 2026, but provides no specific details about any model, product, or feature announcement.

### Google Research — Unlocking dependable responses with Gemini Enterprise Agent Platform’s Agentic RAG

- Company: Google Research (research.google)
- Announced: 2026-06-05T11:26:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/unlocking-dependable-responses-with-gemini-enterprise-agent-platforms-agentic-rag/
- Record: https://forck.live/items/1335-unlocking-dependable-responses-with-gemini-enterprise-agent-platform-s-agentic
- Subject: Research

Google Research introduces a new agentic RAG framework that breaks down complex enterprise queries, iteratively searches for sufficient context, and includes a Sufficient Context Agent to ensure completeness before generating a response. Compared to standard RAG, it increases accuracy on factuality datasets by up to 34% and improves grounding and reasoning accuracy on domain-specific tasks. The framework is hosted on the Gemini Enterprise Agent Platform as Cross-Corpus Retrieval powered by Agentic RAG.

### Manus — One Prompt to Search, Update, and Report Across Your Entire Airtable Base

- Company: Manus (manus.im)
- Announced: 2026-06-05
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://manus.im/blog/deep-dive-airtable-connector
- Record: https://forck.live/items/14716-one-prompt-to-search-update-and-report-across-your-entire-airtable-base
- Subject: Manus

Manus introduced an Airtable connector that lets users manage their Airtable bases through natural language prompts, including querying, updating, and reporting on records. The connector uses Personal Access Tokens for scoped access, and Manus can combine Airtable data with web research and presentation tools for multi-step workflows.

### Suno — Your Voice, Reimagined

- Company: Suno (suno.com)
- Announced: 2026-06-05
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://suno.com/blog/6-tips-for-voices
- Record: https://forck.live/items/10324-your-voice-reimagined
- Subject: Music / Studio

Suno published a guide with six tips for using its Voices feature, including recording in a quiet space, practicing, and providing a voice reference of at least one minute for better results. Voices is available on Web, iOS, and Android, with free and paid plans.

### Amp — Faster Deep & Rush

- Company: Amp (ampcode.com)
- Announced: 2026-06-05
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/faster-deep-rush
- Record: https://forck.live/items/7738-faster-deep-rush
- Subject: Amp

Amp announced speed improvements in its deep and rush modes, with first token arrival 87% faster and entire responses 32% faster at p50, and up to 40% end-to-end speedup on long-horizon tasks, attributed to using websockets for OpenAI communication and a recent rebuild.

### Cursor — Design Mode Improvements

- Company: Cursor (cursor.com)
- Announced: 2026-06-05
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/design-mode-improvements
- Record: https://forck.live/items/2599-design-mode-improvements
- Subject: Cursor

Cursor's Design Mode adds multi-select elements and voice input, allowing users to click, draw, or describe changes by voice to update UI.

### Replit — Build a custom Shopify storefront on Replit

- Company: Replit (replit.com)
- Announced: 2026-06-04T20:55:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replit.com/blog/create-a-custom-shopify-store
- Record: https://forck.live/items/17898-build-a-custom-shopify-storefront-on-replit
- Subject: Replit Agent

Replit announced that its Agent can now build a custom Shopify storefront from a user's description in a single conversation. The Agent generates the front end, creates a new Shopify store, and adds products; the user then claims the store in Shopify, sets up payments, and deploys via Replit. The integration aims to let users design, build, and operate a storefront entirely through conversation with the Agent.

### Google Research — Towards passive heart health monitoring via smartphone camera

- Company: Google Research (research.google)
- Announced: 2026-06-04T19:47:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/towards-passive-heart-health-monitoring-via-smartphone-camera/
- Record: https://forck.live/items/1336-towards-passive-heart-health-monitoring-via-smartphone-camera
- Subject: Research

Google Research introduces a passive heart rate monitoring system (PHRM) that uses a smartphone's front-facing camera to measure heart rate and resting heart rate from facial video during everyday use. The system achieves industry accuracy standards across all skin tones. A pre-trained model called PHRM-mini and a diverse dataset are released for research purposes, accessible to qualified researchers.

### Supabase — Supabase Series F

- Company: Supabase (supabase.com)
- Announced: 2026-06-04T19:00:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/supabase-series-f
- Record: https://forck.live/items/21283-supabase-series-f
- Subject: Supabase

Supabase announced a $500M Series F funding round led by GIC, at a $10B pre-money valuation, with participation from all existing investors, a second investment from Stripe, and new investors Georgian and Salesforce Ventures. The company also released Multigres v0.1 alpha, an open-source scalable operating system for Postgres, and reported that database launches on Supabase have grown 600% in the past year, with nearly 10 million developers now building on the platform.

### NVIDIA — Nemotron 3.5 Content Safety: Customizable Multimodal Safety for Global Enterprise AI

- Company: NVIDIA (nvidia.com)
- Announced: 2026-06-04T18:57:45+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://huggingface.co/blog/nvidia/nemotron-3-5-content-safety
- Record: https://forck.live/items/1489-nemotron-3-5-content-safety-customizable-multimodal-safety-for-global

NVIDIA released Nemotron 3.5 Content Safety, a single 4B-parameter model built on Google Gemma 3 4B IT that unifies multimodal input, multilingual coverage across 12 explicitly trained languages plus zero-shot generalization to ~140 languages, custom enterprise policy enforcement, and optional auditable reasoning traces. The model accepts a user prompt, optional image, and optional assistant response in one context window and produces a safety verdict. NVIDIA also released the multimodal, multilingual safety dataset used to train the model.

### Sakana AI — Introducing Sakana AI’s Recursive Self-Improvement (RSI) Lab

- Company: Sakana AI (sakana.ai)
- Announced: 2026-06-04T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://sakana.ai/rsi-lab
- Record: https://forck.live/items/4682-introducing-sakana-ai-s-recursive-self-improvement-rsi-lab
- Subject: Sakana models

Sakana AI announced the formal establishment of the RSI Lab, a dedicated research group focused on recursive self-improvement technology using foundation models.

### Harvey — Harvey's PitchBook Integration: Private Capital Markets Data, in Harvey

- Company: Harvey (harvey.ai)
- Announced: 2026-06-04T14:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/harvey-pitchbook-integration
- Record: https://forck.live/items/17555-harvey-s-pitchbook-integration-private-capital-markets-data-in-harvey
- Subject: Harvey

Harvey integrated PitchBook's private capital markets data into its platform, allowing users to query company financials, deal history, and market benchmarks directly within Harvey. The feature is cited back to PitchBook and will be available to Harvey customers starting in June.

### ElevenLabs — Introducing Flows Agent in ElevenCreative

- Company: ElevenLabs (elevenlabs.io)
- Announced: 2026-06-04T12:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://elevenlabs.io/blog/introducing-flows-agent
- Record: https://forck.live/items/3352-introducing-flows-agent-in-elevencreative
- Subject: Voice AI

ElevenLabs introduced Flows Agent within ElevenCreative.

### OpenAI — How Endava is redesigning software delivery around AI agents

- Company: OpenAI (openai.com)
- Announced: 2026-06-04T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/endava-frontiers
- Record: https://forck.live/items/83-how-endava-is-redesigning-software-delivery-around-ai-agents
- Subject: GPT / ChatGPT / API

Endava is using OpenAI's AI agents, ChatGPT Enterprise, and Codex to accelerate software delivery, automate workflows, and build an AI-native culture.

### OpenAI — Dreaming: Better memory for a more helpful ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-06-04T09:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/chatgpt-memory-dreaming
- Record: https://forck.live/items/84-dreaming-better-memory-for-a-more-helpful-chatgpt
- Subject: GPT / ChatGPT / API

ChatGPT introduces a new memory system to better remember user preferences and maintain context across conversations.

### Black Forest Labs — FLUX.2 is now on device: ASUS ProArt laptops now support Klein models

- Company: Black Forest Labs (bfl.ai)
- Announced: 2026-06-04T00:06:00+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://bfl.ai/blog/flux2-klein-on-device
- Record: https://forck.live/items/3388-flux-2-is-now-on-device-asus-proart-laptops-now-support-klein-models
- Subject: FLUX

FLUX.2 is now available on device, with ASUS ProArt laptops supporting Klein models.

### Supabase — Multigres v0.1 Alpha: an operating system for Postgres

- Company: Supabase (supabase.com)
- Announced: 2026-06-04
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/multigres-v0-1-alpha
- Record: https://forck.live/items/21284-multigres-v0-1-alpha-an-operating-system-for-postgres
- Subject: Supabase

Supabase released Multigres v0.1 alpha, an open-source operating system for Postgres that provides connection pooling, automatic failover, backup orchestration, and a Kubernetes operator. The alpha release does not yet include sharding, is not production-ready, and future versions may not be backward compatible.

### H Company — H joins the NVIDIA Nemotron Coalition

- Company: H Company (hcompany.ai)
- Announced: 2026-06-04
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://hcompany.ai/newsroom/h-joins-nemotron-coalition
- Record: https://forck.live/items/18368-h-joins-the-nvidia-nemotron-coalition
- Subject: Holo

H Company joined the NVIDIA Nemotron Coalition, an international initiative of open-source AI players announced at NVIDIA's GTC conference in Taipei. The coalition aims to advance open, frontier-level foundation models by combining expertise, computing power, and data resources.

### Manus — This Pharmacist Created A Fitness Tracker that Adapts As Your Body Changes

- Company: Manus (manus.im)
- Announced: 2026-06-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://manus.im/blog/alyak-tv
- Record: https://forck.live/items/14717-this-pharmacist-created-a-fitness-tracker-that-adapts-as-your-body-changes
- Subject: Manus

Yim Yoon-seok, a pharmacist and bodybuilding nutrition coach, used the Manus AI agent to build a fitness tracking web application called Alyak Coach without any prior coding experience. The app features an AI-powered TDEE calculator that adapts weekly based on user data, and within two months it grew to around 300 active users, later reaching 3,900 registered users and generating revenue. An external developer had estimated the project would require six months of full-time work, but Yoon-seok built it in weeks using Manus.

### Amp — Opus 4.8

- Company: Amp (ampcode.com)
- Announced: 2026-06-04
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://ampcode.com/news/opus-4.8
- Record: https://forck.live/items/7740-opus-4-8
- Subject: Amp

Amp released Opus 4.8, a new model that powers its smart mode, replacing Opus 4.7. It is a better coding agent with improved faithfulness, tighter changes, and better self-checking. In internal evals, it solved 62% of tasks, up from 52% for Opus 4.7, and it offers a fast mode at roughly 2.5× speed at 2× base tokens.

### Amp — Agents, Everywhere

- Company: Amp (ampcode.com)
- Announced: 2026-06-04
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/agents-everywhere
- Record: https://forck.live/items/7739-agents-everywhere
- Subject: Amp

Amp announced a new UI and sidebar for monitoring and controlling all Amp agents across web, mobile, and CLI, built on a recently rebuilt distributed system with durable execution and a plugin API.

### Cognition — AI should earn its keep: Introducing the AI Productivity Guarantee

- Company: Cognition (cognition.com)
- Announced: 2026-06-04
- Category: pricing-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cognition.com/blog/ai-guarantee
- Record: https://forck.live/items/7655-ai-should-earn-its-keep-introducing-the-ai-productivity-guarantee
- Subject: Devin / SWE

Cognition introduced an AI Productivity Guarantee for its enterprise customers: if Devin delivers less engineering value than the customer pays for, Cognition will fund usage up to $10M until it does. The guarantee is based on an estimator that measures productive engineering output.

### Cognition — Estimating the Productivity of an Autonomous AI Software Engineer

- Company: Cognition (cognition.com)
- Announced: 2026-06-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://cognition.com/blog/ai-productivity
- Record: https://forck.live/items/7654-estimating-the-productivity-of-an-autonomous-ai-software-engineer
- Subject: Devin / SWE

Cognition developed an automated system to estimate how many productive engineering hours each Devin session is worth, achieving a model with an r_log of 0.74 and showing no bias. The system is now running with customers.

### Cohere — Unpacking AI safety for enterprises

- Company: Cohere (cohere.com)
- Announced: 2026-06-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://cohere.com/blog/unpacking-ai-safety
- Record: https://forck.live/items/4229-unpacking-ai-safety-for-enterprises
- Subject: Command

Cohere published a guide explaining AI safety for enterprises, breaking it down into seven themes based on algorithmic fairness principles, including types of fairness such as allocational and representational harms.

### Cursor — Canvas Design Mode and Context Usage Report

- Company: Cursor (cursor.com)
- Announced: 2026-06-04
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/canvas-improvements
- Record: https://forck.live/items/2601-canvas-design-mode-and-context-usage-report
- Subject: Cursor

Cursor introduces Design Mode for faster canvas editing, a context usage report in canvas, and other quality-of-life improvements such as full-screen shared canvases, embeddable buttons, and better error handling.

### Cursor — Custom stores, custom tools, and auto-review for the Cursor SDK

- Company: Cursor (cursor.com)
- Announced: 2026-06-04
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/sdk-updates-jun-2026
- Record: https://forck.live/items/2600-custom-stores-custom-tools-and-auto-review-for-the-cursor-sdk
- Subject: Cursor

Cursor released a batch of new functionality across its TypeScript and Python SDKs, including custom tools, auto-review, JSONL and custom stores, nested subagents, and various reliability, performance, and platform improvements.

### Hugging Face — Designing the hf CLI as an agent-optimized way to work with the Hub

- Company: Hugging Face (huggingface.co)
- Announced: 2026-06-04
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/hf-cli-for-agents
- Record: https://forck.live/items/1490-designing-the-hf-cli-as-an-agent-optimized-way-to-work-with-the-hub
- Subject: Platform

Hugging Face redesigned its hf CLI to provide agent-optimized output, auto-detecting coding agents like Claude Code and Codex to render TSV format instead of human-friendly tables, reducing token usage by up to 6x on complex multi-step tasks.

### OpenAI — Biodefense in the Intelligence Age

- Company: OpenAI (openai.com)
- Announced: 2026-06-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/biodefense-in-the-intelligence-age
- Record: https://forck.live/items/85-biodefense-in-the-intelligence-age
- Subject: GPT / ChatGPT / API

OpenAI published an action plan for AI-powered biological resilience.

### Replit — Meet Replit SEO Agent

- Company: Replit (replit.com)
- Announced: 2026-06-03T20:01:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/seo-agent
- Record: https://forck.live/items/17899-meet-replit-seo-agent
- Subject: Replit Agent

Replit introduced SEO Agent, a tool that scans published apps for discoverability issues and can apply fixes. It lives in the new Growth dashboard alongside traffic monitoring. Replit also upgraded Agent defaults to include semantic HTML, meta tags, Open Graph previews, and sitemap/robots files.

### Google Research — The next chapter in flood resilience: Open sourcing Google’s hydrology framework

- Company: Google Research (research.google)
- Announced: 2026-06-03T18:37:00+00:00
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/the-next-chapter-in-flood-resilience-open-sourcing-googles-hydrology-framework/
- Record: https://forck.live/items/1337-the-next-chapter-in-flood-resilience-open-sourcing-google-s-hydrology-framework
- Subject: Research

Google Research has open-sourced its hydrology modeling framework on GitHub, which includes two versions of the hydrological model: the original version and an upgraded v2 model using ME-LSTM architecture. The upgraded model extends the reliable predictive horizon by six days in gauged basins and one day in ungauged basins compared to the previous version. The framework allows researchers and forecasters to train AI flood forecasting models with the same architecture as Google's Flood Hub.

### Meta — Lights Out, Systems On: Validating Instant Power Loss Readiness

- Company: Meta (meta.com)
- Announced: 2026-06-03T17:00:44+00:00
- Category: infrastructure-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://engineering.fb.com/2026/06/03/data-center-engineering/lights-out-systems-on-validating-instant-power-loss-readiness/
- Record: https://forck.live/items/1428-lights-out-systems-on-validating-instant-power-loss-readiness
- Subject: Llama / infrastructure

Meta introduces Instantaneous PowerLoss Storm, a new testing paradigm within its infrastructure for handling and mitigating instant or zero-notice power loss in data centers, built from the ground up into the DC stack.

### OpenAI — Introducing new capabilities to GPT-Rosalind

- Company: OpenAI (openai.com)
- Announced: 2026-06-03T13:15:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/introducing-new-capabilities-to-gpt-rosalind
- Record: https://forck.live/items/86-introducing-new-capabilities-to-gpt-rosalind
- Subject: GPT / ChatGPT / API

OpenAI announced new capabilities for GPT-Rosalind in life sciences research, including enhanced biological reasoning, medicinal chemistry expertise, genomics analysis, and experimental workflow capabilities.

### Harvey — Measure and Manage Collaboration in Shared Spaces

- Company: Harvey (harvey.ai)
- Announced: 2026-06-03T13:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/shared-spaces-analytics-and-management
- Record: https://forck.live/items/17556-measure-and-manage-collaboration-in-shared-spaces
- Subject: Harvey

Harvey introduced Shared Spaces Analytics, External Collaboration Management, and a redesigned Share Modal for its legal collaboration platform. The analytics dashboard provides real-time metrics on active users, queries, workflow executions, and review table activity, filterable by date, organization, user, and resource. External Collaboration Management gives admins a dedicated view of externally shared Spaces with activity logs and permission controls.

### Hugging Face — Direct Preference Optimization Beyond Chatbots

- Company: Hugging Face (huggingface.co)
- Announced: 2026-06-03T12:55:11+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/Dharma-AI/direct-preference-optimization-beyond-chatbots
- Record: https://forck.live/items/1491-direct-preference-optimization-beyond-chatbots
- Subject: Platform

Using Direct Preference Optimization (DPO) on OCR models to reduce text degeneration, using rejection pairs from the model's own failures. It reports an average reduction of 59.4% in degeneration rate across model families.

### OpenAI — How Wasmer used Codex to build a Node.js runtime for the edge

- Company: OpenAI (openai.com)
- Announced: 2026-06-03T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/wasmer
- Record: https://forck.live/items/87-how-wasmer-used-codex-to-build-a-node-js-runtime-for-the-edge
- Subject: GPT / ChatGPT / API

Wasmer used Codex with GPT-5.5 to build a Node.js runtime for the edge, accelerating development 10x to 20x and shipping in weeks instead of months.

### Lovable — How one data scientist enabled a 150-person organization with a single Lovable app

- Company: Lovable (lovable.dev)
- Announced: 2026-06-03T10:05:15+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lovable.dev/blog/how-one-data-scientist-enabled-a-150-person-org-with-a-single-lovable-app
- Record: https://forck.live/items/12076-how-one-data-scientist-enabled-a-150-person-organization-with-a-single-lovable
- Subject: Lovable / AI app builder

A data scientist at Lovable built an internal app called Lovsight during a hackathon using Lovable's platform. The app integrates data from multiple sources via APIs and became used by roughly 90% of the 150-person organization within weeks, without writing code. It now serves as key company infrastructure for analytics across teams.

### OpenAI — OpenAI public policy agenda

- Company: OpenAI (openai.com)
- Announced: 2026-06-03T10:00:00+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/public-policy-agenda
- Record: https://forck.live/items/89-openai-public-policy-agenda
- Subject: GPT / ChatGPT / API

OpenAI outlines its public policy agenda covering safety, youth protection, workforce transition, and global standards to ensure AI benefits society.

### OpenAI — A blueprint for democratic governance of frontier AI

- Company: OpenAI (openai.com)
- Announced: 2026-06-03T10:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/frontier-safety-blueprint
- Record: https://forck.live/items/88-a-blueprint-for-democratic-governance-of-frontier-ai
- Subject: GPT / ChatGPT / API

OpenAI proposes a federal framework for governing frontier AI in the U.S., focusing on safety, resilience, and national security.

### OpenClaw — Skill Workshop: Turn Agent Work Into Reusable Skills

- Company: OpenClaw (openclaw.ai)
- Announced: 2026-06-03T06:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openclaw.ai/blog/openclaw-agent-skill-workshop/
- Record: https://forck.live/items/17762-skill-workshop-turn-agent-work-into-reusable-skills
- Subject: OpenClaw

OpenClaw introduced Skill Workshop, a feature that adds a review step before an agent's skill creation or revision becomes active. Skills are reusable procedures that change future agent behavior, and Skill Workshop ensures they start as non-active proposals that users can inspect, revise, apply, or reject. The same proposal flow works across chat, the Control UI, channels, the CLI, and Gateway.

### Black Forest Labs — Martin Scorsese is an advisor to Black Forest Labs

- Company: Black Forest Labs (bfl.ai)
- Announced: 2026-06-03T01:22:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://bfl.ai/blog/martin-scorsese-bfl-advisor
- Record: https://forck.live/items/3389-martin-scorsese-is-an-advisor-to-black-forest-labs
- Subject: FLUX

Martin Scorsese has been announced as an advisor to Black Forest Labs.

### Ideogram — Ideogram 4.0 Technical Details: Open model at the forefront of design

- Company: Ideogram (ideogram.ai)
- Announced: 2026-06-03
- Category: open-weight-release
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://ideogram.ai/blog/ideogram-4.0
- Record: https://forck.live/items/15464-ideogram-4-0-technical-details-open-model-at-the-forefront-of-design
- Subject: Ideogram image models

Ideogram released Ideogram 4.0, a 9.3B parameter open-weight text-to-image model. The model uses a 34-layer Diffusion Transformer with a Qwen3-VL-8B-Instruct text encoder and is trained exclusively on structured JSON captions with per-element styling, optional bounding boxes, and color palettes. The reference inference pipeline validates prompts against the JSON schema before generation.

### Krea — Krea 2 Turbo: Generate Images in 2 Seconds

- Company: Krea (krea.ai)
- Announced: 2026-06-03
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.krea.ai/blog/krea-2-turbo
- Record: https://forck.live/items/15074-krea-2-turbo-generate-images-in-2-seconds
- Subject: Krea 2 / AI creative platform

Krea released Krea 2 Turbo, a faster version of its Krea 2 image generation model that produces images in 2 seconds while retaining compatibility with style references, moodboards, and LoRAs. The company notes that Krea 2 Medium and Krea 2 Large remain better choices when exact style adherence or final-image polish is critical.

### Suno — The Next Chapter for Suno

- Company: Suno (suno.com)
- Announced: 2026-06-03
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://suno.com/blog/series-d-announcement
- Record: https://forck.live/items/10325-the-next-chapter-for-suno
- Subject: Music / Studio

Suno announced it has raised over $400 million in Series D funding at a $5.4 billion post-money valuation. The round was led by Bond Capital with participation from several other investors. The company plans to use the funding to accelerate its mission of helping people express themselves through music and to roll out its first music model developed in partnership with the music industry.

### Luma — Human, after all.

- Company: Luma (lumalabs.ai)
- Announced: 2026-06-03
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lumalabs.ai/news/human-after-all
- Record: https://forck.live/items/7963-human-after-all
- Subject: Ray / Dream Machine

Luma has joined Human After All, a new AI Creator Studio launched by Webedia-Elephant, as a creative partner alongside Google and ElevenLabs. The studio focuses on AI-assisted creation, offering creators access to Luma's tools and in-person support in Paris through the summer.

### World Labs — A Functional Taxonomy of World Models

- Company: World Labs (worldlabs.ai)
- Announced: 2026-06-03
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.worldlabs.ai/blog/taxonomy-of-world-models
- Record: https://forck.live/items/7341-a-functional-taxonomy-of-world-models
- Subject: Marble / Atlas

World Labs introduces a functional taxonomy of world models, categorizing them into three types: renderers, simulators, and planners, each serving distinct roles in AI's understanding of space and time.

### xAI — Grok Becomes the Voice of Vapi

- Company: xAI (x.ai)
- Announced: 2026-06-03
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/grok-vapi
- Record: https://forck.live/items/3459-grok-becomes-the-voice-of-vapi
- Subject: Grok

Grok is now integrated with Vapi, providing voice capabilities to Vapi agents.

### xAI — Grok Imagine 1.5 Preview

- Company: xAI (x.ai)
- Announced: 2026-06-03
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://x.ai/news/grok-imagine-1-5
- Record: https://forck.live/items/3458-grok-imagine-1-5-preview
- Subject: Grok

xAI released a preview of its image-to-video model grok-imagine-video-1.5-preview via the xAI API.

### Cohere — Co/plot: Supporting the research process through visualization

- Company: Cohere (cohere.com)
- Announced: 2026-06-03
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://cohere.com/blog/coplot
- Record: https://forck.live/items/3280-co-plot-supporting-the-research-process-through-visualization
- Subject: Command

A blog post titled 'Co/plot: Supporting the research process through visualization' with a date and read time, but provides no further details about the product, model, or capabilities.

### Cursor — Organizations for Cursor Enterprise

- Company: Cursor (cursor.com)
- Announced: 2026-06-03
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/enterprise-organizations
- Record: https://forck.live/items/2602-organizations-for-cursor-enterprise
- Subject: Cursor

Cursor Enterprise now allows customers to manage multiple teams under a single organization, with separate security, governance, budget, and feature controls for each team, plus groups for lightweight user collections. These capabilities are generally available to all Enterprise customers.

### Hugging Face — Adding MCP Tools to Reachy Mini

- Company: Hugging Face (huggingface.co)
- Announced: 2026-06-03
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/adding-mcp-tools-to-reachy-mini
- Record: https://forck.live/items/1492-adding-mcp-tools-to-reachy-mini
- Subject: Platform

Hugging Face announced that Reachy Mini's conversation app can now use remote tools hosted in public Hugging Face Spaces, called over MCP. This adds a third type of tool alongside built-in and custom local tools, allowing capabilities like weather and web search to be published, shared, and updated independently from the robot app itself. The feature is accessible via a new command-line interface (`reachy-mini-conversation-app tool-spaces add`) and works by discovering remote tools and enabling them per profile through `tools.txt`.

### H Company — Holo3.1: Fast & Local Computer Use Agents

- Company: H Company (hcompany.ai)
- Announced: 2026-06-02T14:13:23+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://huggingface.co/blog/Hcompany/holo31
- Record: https://forck.live/items/1493-holo3-1-fast-local-computer-use-agents

Hcompany released the Holo3.1 family of computer-use models, improving robustness across environments (web, desktop, mobile), agent frameworks, and deployment targets. New model sizes include 0.8B, 4B, 9B, and 35B-A3B. For the first time, quantized checkpoints (FP8, Q4 GGUF, NVFP4) are released for local inference. Performance gains are reported on AndroidWorld and OSWorld benchmarks.

### Harvey — Move From Opinions to Final Outputs With US Case Law

- Company: Harvey (harvey.ai)
- Announced: 2026-06-02T13:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/us-case-law-source
- Record: https://forck.live/items/17557-move-from-opinions-to-final-outputs-with-us-case-law
- Subject: Harvey

Harvey added a US Case Law knowledge source that gives lawyers direct access to over 9 million court opinions for research, analysis, and drafting within the platform. The feature includes direct citations, the ability to view case content alongside results, and integration with Harvey's Word Add-In and Agent Builder. It is available starting June 3rd on a rolling basis to Harvey customers.

### OpenAI — Travelers deploys AI-powered claims countrywide with OpenAI

- Company: OpenAI (openai.com)
- Announced: 2026-06-02T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/travelers
- Record: https://forck.live/items/90-travelers-deploys-ai-powered-claims-countrywide-with-openai
- Subject: GPT / ChatGPT / API

Travelers has deployed an AI-powered Claim Assistant built with OpenAI to assist customers with filing claims, provide 24/7 support, and scale operations during peak demand.

### OpenAI — ChatGPT Sites

- Company: OpenAI (openai.com)
- Announced: 2026-06-02T10:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/chatgpt-sites
- Record: https://forck.live/items/91-chatgpt-sites
- Subject: GPT / ChatGPT / API

The source text suggests using Codex to build lightweight websites or apps, but does not announce any new model, product, or capability. It is a generic guidance statement without a launch or change.

### OpenAI — Codex for every role, tool, and workflow

- Company: OpenAI (openai.com)
- Announced: 2026-06-02T09:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/codex-for-every-role-tool-workflow
- Record: https://forck.live/items/92-codex-for-every-role-tool-and-workflow
- Subject: GPT / ChatGPT / API

OpenAI announces new Codex plugins, sites, and annotations designed to assist various professional roles such as analysts, marketers, designers, and investors in leveraging AI more effectively.

### OpenAI — Advancing youth safety and opportunity through global leadership

- Company: OpenAI (openai.com)
- Announced: 2026-06-02T07:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/advancing-youth-safety-and-opportunity-through-global-leadership
- Record: https://forck.live/items/93-advancing-youth-safety-and-opportunity-through-global-leadership
- Subject: GPT / ChatGPT / API

OpenAI proposes an international institute to strengthen safeguards, standards, and opportunities for young people regarding AI safety.

### OpenAI — Codex is becoming a productivity tool for everyone

- Company: OpenAI (openai.com)
- Announced: 2026-06-02T02:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/codex-for-knowledge-work
- Record: https://forck.live/items/94-codex-is-becoming-a-productivity-tool-for-everyone
- Subject: GPT / ChatGPT / API

OpenAI published a report titled 'The Next Era of Knowledge Work' that examines how Codex is enhancing productivity through AI-powered research, data analysis, workflow automation, and content creation.

### Gradium — Semantic VAD: turn detection that uses meaning, not silence

- Company: Gradium (gradium.ai)
- Announced: 2026-06-02
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://gradium.ai/blog/semantic-vad
- Record: https://forck.live/items/18317-semantic-vad-turn-detection-that-uses-meaning-not-silence
- Subject: Gradium TTS / Phonon

Gradium's speech-to-text system now includes semantic VAD (voice activity detection) that predicts turn completion from linguistic content rather than silence alone. The model emits inactivity probabilities at multiple future horizons every 80 milliseconds, allowing agents to avoid interrupting users during hesitations or pauses mid-sentence. The feature is designed to work within the same WebSocket stream as transcription, adding no extra latency.

### Harvey — Harvey Opens in Singapore

- Company: Harvey (harvey.ai)
- Announced: 2026-06-02
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/harvey-opens-in-singapore
- Record: https://forck.live/items/17558-harvey-opens-in-singapore
- Subject: Harvey

Harvey announced the opening of a new office in Singapore to support its growing customer base in the Asia Pacific region. The office will house teams in legal engineering, customer success, sales, and G&A. Harvey also partnered with the National University of Singapore Faculty of Law to advance legal AI education.

### Microsoft AI — Introducing MAI-Code-1-Flash | Microsoft AI

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/introducingmai-code-1-flash/
- Record: https://forck.live/items/10840-introducing-mai-code-1-flash-microsoft-ai
- Subject: MAI

Microsoft introduced MAI-Code-1-Flash, a coding model trained from scratch on clean, enterprise-grade data without distillation from third-party models. The model is rolling out to GitHub Copilot individual users in Visual Studio Code and outperforms Claude Haiku 4.5 on all tested coding benchmarks, including a +16-point lead on SWE-Bench Pro (51.2% vs. 35.2%). It is designed for agentic coding in real developer environments and uses adaptive solution length control to reduce token usage by up to 60% on harder problems.

### Microsoft AI — Introducing MAI-Transcribe-1.5 | Microsoft AI

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/mai-transcribe-1-5more-accurate-context-aware-and-built-for-production/
- Record: https://forck.live/items/10839-introducing-mai-transcribe-1-5-microsoft-ai
- Subject: MAI

Microsoft AI released MAI-Transcribe-1.5, a multilingual speech-to-text model covering 43 languages with a best-in-class Word Error Rate on the FLEURS benchmark. The model includes keyword biasing that improves recognition of domain-specific terms by up to 30% on FLEURS, and can transcribe an hour of audio in under 15 seconds. It is being integrated into Copilot, Teams, GitHub, and Dynamics 365 Contact Centre, and is available via Foundry.

### Microsoft AI — Building a hill-climbing machine: Launching seven new MAI models | Microsoft AI

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://microsoft.ai/news/building-a-hillclimbing-machine-launching-seven-new-mai-models/
- Record: https://forck.live/items/10811-building-a-hill-climbing-machine-launching-seven-new-mai-models-microsoft-ai
- Subject: MAI

Microsoft AI announced seven new models spanning image, voice, transcription, coding, and reasoning capabilities. MAI-Thinking-1 is positioned as a flagship reasoning model matching leading models on software engineering benchmarks; MAI-Code-1-Flash is an inference-efficient coding model with 5 billion active parameters integrated into GitHub Copilot; MAI-Image-2.5 supports text-to-image and image editing; MAI Transcribe-1.5 is described as the best transcription model with five times faster performance than competitors across 43 languages; MAI-Voice-2 provides speech generation across 15 languages with voice adaptation capabilities. The models will be available on Foundry, OpenRouter, Fireworks, and Baseten, with developers able to tune model weights. Microsoft also announced a collaboration with Mayo Clinic to co-create a frontier AI model for healthcare.

### Microsoft AI — Microsoft Build 2026: MAI keynote transcript | Microsoft AI

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/microsoft-build-2026-mai-keynote-transcript/
- Record: https://forck.live/items/10810-microsoft-build-2026-mai-keynote-transcript-microsoft-ai
- Subject: MAI

Microsoft AI announced a family of seven new MAI models at Build 2026, including image, voice, transcription, thinking, and coding models. MAI-Thinking-1 is a 35B active parameter MoE with a 256K context window, achieving 97% on AIME 25 and 53% on SWE Bench Pro. MAI-Code-1-Flash has 5B parameters and achieves 51% on SWE Bench Pro. The models are available on Foundry, Open Router, Fireworks, and Baseten, and are integrated into Microsoft products like PowerPoint, Copilot, Teams, GitHub, and VS Code.

### Microsoft AI — Our values in operation: Health | Microsoft AI

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://microsoft.ai/news/our-values-in-operation-health/
- Record: https://forck.live/items/10787-our-values-in-operation-health-microsoft-ai
- Subject: MAI

Microsoft AI describes its values and approach to health AI, citing tens of millions of daily health questions across its consumer products and a recent analysis of half a million Copilot conversations. It highlights its Diagnostic Orchestrator (MAI DxO) research and a strategic collaboration with Mayo Clinic to develop a frontier AI model for healthcare, which will be owned by Mayo Clinic and first deployed within Mayo Clinic. The post also notes the preview launch of Copilot Health for Microsoft 365 subscribers in the US, which integrates health records, wearable data, and health history.

### Microsoft AI — MAI-Image-2.5 launches at No. 2 for image editing on Arena | Microsoft AI

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/introducing-mai-image-2-5/
- Record: https://forck.live/items/10786-mai-image-2-5-launches-at-no-2-for-image-editing-on-arena-microsoft-ai
- Subject: MAI

Microsoft AI released MAI-Image-2.5 and MAI-Image-2.5-Flash, image generation and editing models that rank No. 2 on Arena's image-editing leaderboard. MAI-Image-2.5 offers improved text-to-image quality, complex visual reasoning, fine-grained edit control, and face identity consistency. The models are available to developers in Foundry and are rolling out to PowerPoint and OneDrive.

### Microsoft AI — Introducing MAI-Voice-2 | Microsoft AI

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/mai-voice-2/
- Record: https://forck.live/items/10767-introducing-mai-voice-2-microsoft-ai
- Subject: MAI

Microsoft AI launched MAI-Voice-2, a text-to-speech model supporting 15 languages with granular emotion control, zero-shot voice prompting, and code-switching capabilities. The model is preferred over its predecessor 72% of the time and is now available in Microsoft Foundry, with integration into VSCode and Dynamics 365 Contact Center.

### Microsoft AI — Introducing MAI-Code-1-Flash

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/introducingmai-code-1-flash
- Record: https://forck.live/items/8428-introducing-mai-code-1-flash
- Subject: MAI

Microsoft AI introduced MAI-Code-1-Flash, a new coding model built for fast, efficient assistance in developer workflows, trained on enterprise-grade data without distillation from third-party models. It is rolling out to GitHub Copilot individual users in Visual Studio Code. The model outperforms Claude Haiku 4.5 on coding benchmarks and uses adaptive solution length control for efficiency.

### Microsoft AI — Building a hill-climbing machine: Launching seven new MAI models

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://microsoft.ai/news/building-a-hillclimbing-machine-launching-seven-new-mai-models
- Record: https://forck.live/items/8427-building-a-hill-climbing-machine-launching-seven-new-mai-models
- Subject: MAI

Microsoft AI announced a family of seven new models: MAI-Thinking-1, MAI-Code-1-Flash, MAI-Image-2.5, MAI-Transcribe-1.5, MAI-Voice-2, and MAI-Voice-2-Flash. These models span reasoning, coding, image, transcription, and voice capabilities. The company also introduced Frontier Tuning for model adaptation and announced a collaboration with Mayo Clinic to co-create a frontier AI model for healthcare.

### Microsoft AI — Introducing MAI-Transcribe-1.5

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/mai-transcribe-1-5more-accurate-context-aware-and-built-for-production
- Record: https://forck.live/items/8426-introducing-mai-transcribe-1-5
- Subject: MAI

Microsoft AI launched MAI-Transcribe-1.5, a multilingual speech-to-text model with best-in-class Word-Error-Rate across 43 languages, expanded from 25. The model includes keyword biasing and is being integrated into Copilot, Teams, GitHub, and Dynamics 365 Contact Centre, and is available in Foundry. It can transcribe an hour of audio in under 15 seconds.

### Microsoft AI — Microsoft Build 2026: MAI keynote transcript

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/microsoft-build-2026-mai-keynote-transcript
- Record: https://forck.live/items/8425-microsoft-build-2026-mai-keynote-transcript
- Subject: MAI

Microsoft AI announced a family of seven new models covering image, voice, transcription, thinking, and coding. The models are MAI-Image-2.5 and its Flash variant, MAI-Transcribe-1.5, MAI-Voice-2 and Voice-2-Flash, MAI-Thinking-1, and MAI-Code-1-Flash. They are being made available on Foundry and other platforms.

### Microsoft AI — Our values in operation: Health

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://microsoft.ai/news/our-values-in-operation-health
- Record: https://forck.live/items/8424-our-values-in-operation-health
- Subject: MAI

Microsoft AI opened Copilot Health in preview for Microsoft 365 Personal, Family, and Premium subscribers in the United States.

### Microsoft AI — MAI-Image-2.5 launches at No. 2 for image editing on Arena

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/introducing-mai-image-2-5
- Record: https://forck.live/items/8423-mai-image-2-5-launches-at-no-2-for-image-editing-on-arena
- Subject: MAI

Microsoft AI launched MAI-Image-2.5, an image generation and editing model, alongside MAI-Image-2.5-Flash, a faster variant. The models are available to developers in Foundry and power Microsoft products such as PowerPoint and OneDrive. MAI-Image-2.5 ranks No. 2 on Arena's Image Edit leaderboard.

### Microsoft AI — Introducing MAI-Voice-2

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-06-02
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://microsoft.ai/news/mai-voice-2
- Record: https://forck.live/items/8422-introducing-mai-voice-2
- Subject: MAI

Microsoft AI launched MAI-Voice-2, a text-to-speech model supporting 15 languages with emotion control, zero-shot voice prompting, and consent guardrails. It is available in Microsoft Foundry and integrated into VSCode and Dynamics 365 Contact Center. The model is preferred over its predecessor 72% of the time.

### Amp — The End of Public Threads

- Company: Amp (ampcode.com)
- Announced: 2026-06-02
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/end-of-public-threads
- Record: https://forck.live/items/7741-the-end-of-public-threads
- Subject: Amp

Amp removed public, internet-wide, discoverable thread sharing, citing the risk of snippets of sensitive files being included in threads as agents read more files into context. Users can still share threads within a workspace or as Unlisted via an unguessable URL. Existing public threads have been converted to Unlisted, and user profiles no longer show threads.

### Cognition — Introducing Devin Desktop

- Company: Cognition (cognition.com)
- Announced: 2026-06-02
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cognition.com/blog/introducing-devin-desktop
- Record: https://forck.live/items/7656-introducing-devin-desktop
- Subject: Devin / SWE

Cognition introduced Devin Desktop, the next generation of Windsurf, an IDE that integrates agent management for software engineers. Devin Desktop features the Agent Command Center, Spaces for shared context between agents, and support for the Agent Client Protocol (ACP) to run any ACP-compatible agent alongside Devin.

### Together AI — Serving MiniMax-M3 for efficient inference: Unlocking 1M-Token Context and Multimodality Without Regrets

- Company: Together AI (together.ai)
- Announced: 2026-06-02
- Category: api-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.together.ai/blog/serving-minimax-m3-for-efficient-inference-unlocking-1m-token-context-and-multimodality-without-regrets
- Record: https://forck.live/items/2343-serving-minimax-m3-for-efficient-inference-unlocking-1m-token-context-and
- Subject: Inference platform

Together AI announces it is the preferred cloud partner for MiniMax M3 and will host the open-weights model as a developer endpoint upon its public release. The company's Inference and Kernel teams delivered optimizations including a KV-Block-Major sparse attention kernel, paged attention integration for MSA, an index scoring kernel, and a Rust-based multimodal preprocessing gateway, achieving 81–125% throughput improvements across concurrency levels. The model supports a 1M-token context window and native multimodality.

### Replit — Unlocking a New Way to Build Enterprise Data Apps with Microsoft Fabric

- Company: Replit (replit.com)
- Announced: 2026-06-01T17:25:03+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/microsoft-fabric
- Record: https://forck.live/items/17900-unlocking-a-new-way-to-build-enterprise-data-apps-with-microsoft-fabric
- Subject: Replit Agent

Replit announced a collaboration with Microsoft to integrate its AI-powered development platform with Microsoft Fabric. The integration uses Rayfin, an open-source SDK and CLI, to let teams build and deploy production-ready applications directly into Fabric with enterprise governance and security. The partnership aims to compress the time from idea to deployment for enterprise data apps.

### OpenAI — Our views on AI policy and political advocacy

- Company: OpenAI (openai.com)
- Announced: 2026-06-01T17:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/our-views-on-ai-policy-and-political-advocacy
- Record: https://forck.live/items/95-our-views-on-ai-policy-and-political-advocacy
- Subject: GPT / ChatGPT / API

OpenAI outlines its approach to AI policy, political advocacy, transparency, support for thoughtful regulation and AI safety, and states that no outside political group speaks on the company's behalf.

### Lovable — Building apps using TanStack Start

- Company: Lovable (lovable.dev)
- Announced: 2026-06-01T16:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lovable.dev/blog/building-apps-using-tanstack-start
- Record: https://forck.live/items/12077-building-apps-using-tanstack-start
- Subject: Lovable / AI app builder

Lovable announced that starting May 13, new projects are server-side rendered and powered by TanStack Start, a React framework for building full-stack apps. Existing projects keep working as before but gain pre-rendering for improved crawler readability. The change allows server-only logic to reside directly within component files, managed by the build process.

### Hugging Face — Introducing Mellum2: A 12B Mixture-of-Experts Model by JetBrains

- Company: Hugging Face (huggingface.co)
- Announced: 2026-06-01T15:45:17+00:00
- Category: open-weight-release
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://huggingface.co/blog/JetBrains/mellum2-launch
- Record: https://forck.live/items/1494-introducing-mellum2-a-12b-mixture-of-experts-model-by-jetbrains
- Subject: Platform

JetBrains releases Mellum2, a 12B-parameter Mixture-of-Experts model trained from scratch on text and code, with 2.5B active parameters per token, under Apache 2.0 license. It is optimized for low-latency inference and targets routing, RAG, sub-agents, and private deployment.

### Harvey — Why we Built our own Cloud Agent Infrastructure

- Company: Harvey (harvey.ai)
- Announced: 2026-06-01T14:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/why-we-built-our-own-cloud-agent-infrastructure
- Record: https://forck.live/items/17559-why-we-built-our-own-cloud-agent-infrastructure
- Subject: Harvey

Harvey built its own cloud agent infrastructure because existing managed runtimes from frontier labs and cloud providers do not meet law firms' requirements for multi-model support, zero data retention, and cost control. The company argues that single-model lock-in is untenable for firms with conflicted clients, and that zero data retention must be an architectural property rather than a post-hoc deletion feature.

### Hugging Face — Beyond LLMs: Why Scalable Enterprise AI Adoption Depends on Agent Logic

- Company: Hugging Face (huggingface.co)
- Announced: 2026-06-01T13:51:18+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/ibm-research/agent-logic-and-scalable-ai-adoption
- Record: https://forck.live/items/1495-beyond-llms-why-scalable-enterprise-ai-adoption-depends-on-agent-logic
- Subject: Platform

IBM Research publishes a blog post on Hugging Face discussing the importance of agent logic for scalable enterprise AI adoption. It describes four agent-based use cases: legacy code understanding with watsonx Code Assistant for Z (using Mistral Medium 250B), test generation with Aster (using Devstral 24B), incident response, and compliance modernization. The post highlights significant token consumption reductions (up to 30× for legacy code, 15× for test generation) compared to frontier LLM-only approaches, without announcing a new product or model release. The category is other because this is a technical article, not a launch or update of a product or model. The models_affected are the two models explicitly named in the text: Mistral Medium 250B and Devstral 24B. No license, context_window, or pricing information is provided in the source text. The evidence_excerpt is a verbatim definition of agent logic from the article. The summary is based solely on the source text, and no prior knowledge or extrapolation is used. The provider is Hugging Face, and the title and source URL are as given. The source text is a blog post, not a product announcement or model release, so category is 'other' per the taxonomy. The evidence_excerpt directly supports the summary by defining agent logic, which is the central concept. The models_affected are included because they are explicitly mentioned in the context of the use cases. All other fields are left empty as they are not stated in the source text. The output is in JSON format as required, with the schema matched. The response is concise and adheres to the instructions: no prior knowledge, no guessing, verbatim evidence, and precise category selection. The source text does not state a context window, license, pricing, or open-weight status, so those are left empty. The JSON is validated against the schema with all required fields present. The is_open_weight field is set to false because the source does not mention open weights, and the field is boolean. The models_affected array contains only the model names that literally appear in the source text. The summary focuses on the key points of the blog post without introducing new information. The evidence_excerpt is a direct quote from the source text that supports the summary. The category is correctly identified as 'other' because the post is a research article, not a product launch, model update, or other specific category. The response complies with all rules.

### ElevenLabs — Homecoming: bringing the ElevenLabs Summit to Warsaw

- Company: ElevenLabs (elevenlabs.io)
- Announced: 2026-06-01T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://elevenlabs.io/blog/warsaw-summit-homecoming
- Record: https://forck.live/items/3353-homecoming-bringing-the-elevenlabs-summit-to-warsaw
- Subject: Voice AI

ElevenLabs announces that it is bringing its Summit event to Warsaw.

### OpenAI — Building the infrastructure for the Intelligence Age in Michigan

- Company: OpenAI (openai.com)
- Announced: 2026-06-01T12:00:00+00:00
- Category: infrastructure-release
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/stargate-michigan-data-center
- Record: https://forck.live/items/96-building-the-infrastructure-for-the-intelligence-age-in-michigan
- Subject: GPT / ChatGPT / API

OpenAI has begun construction on a 1 gigawatt data center in Michigan as part of the Stargate project, aimed at expanding AI infrastructure, creating jobs, and supporting communities.

### Runway — Runway Opens London HQ, Bringing World Model Research Hub to the UK and Europe

- Company: Runway (runway.com)
- Announced: 2026-06-01T10:10:10.110000+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://runway.com/news/runway-opens-london-hq
- Record: https://forck.live/items/7875-runway-opens-london-hq-bringing-world-model-research-hub-to-the-uk-and-europe
- Subject: Gen / Aleph

Runway announced the opening of a London headquarters and a research hub focused on general world models. The company plans to invest $100 million into the UK AI ecosystem over the next 18 months, with the figure more than doubling through 2028. The expansion aims to grow Runway's European research and commercial presence.

### OpenAI — OpenAI frontier models and Codex are now available on AWS

- Company: OpenAI (openai.com)
- Announced: 2026-06-01T10:00:00+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/openai-frontier-models-and-codex-are-now-available-on-aws
- Record: https://forck.live/items/97-openai-frontier-models-and-codex-are-now-available-on-aws
- Subject: GPT / ChatGPT / API

OpenAI frontier models and Codex are now generally available on AWS, enabling enterprises to use OpenAI through AWS environments, controls, and procurement workflows.

### Lovable — How Lovable protects your apps automatically

- Company: Lovable (lovable.dev)
- Announced: 2026-06-01T08:15:56+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/how-lovable-protects-your-apps-automatically
- Record: https://forck.live/items/12078-how-lovable-protects-your-apps-automatically
- Subject: Lovable / AI app builder

Lovable introduced automatic security scans that run before publishing, checking for database misconfigurations, missing RLS policies, and authorization gaps. The basic scan takes 10-15 seconds and can auto-fix low-risk issues. Enterprise users can schedule deep security scans weekly or monthly. A security memory feature reduces ignored findings by about 20% and increases scanning accuracy.

### OpenClaw — OpenClaw Collaborates with NVIDIA for Stronger Agent Skill Security

- Company: OpenClaw (openclaw.ai)
- Announced: 2026-06-01T05:30:00+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openclaw.ai/blog/openclaw-nvidia-skill-security/
- Record: https://forck.live/items/17763-openclaw-collaborates-with-nvidia-for-stronger-agent-skill-security
- Subject: OpenClaw

OpenClaw announced a collaboration with NVIDIA on agent skill security, integrating NVIDIA SkillSpector and Skill Cards into ClawHub's verification pipeline. It released a public dataset of 67,453 security scan outcomes on Hugging Face, finding that three scanners (VirusTotal, static analysis, SkillSpector) flagged different risks with minimal overlap. The ClawScan LLM-as-judge weighs all scanner outputs to produce a final verdict of Clean, Suspicious, or Malicious for each skill.

### Runway — Introducing the Cosmos Coalition

- Company: Runway (runway.com)
- Announced: 2026-06-01T05:30:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://runway.com/news/introducing-cosmos-coalition
- Record: https://forck.live/items/7876-introducing-the-cosmos-coalition
- Subject: Gen / Aleph

Runway announced it is joining the Cosmos Coalition as a founding member, working alongside NVIDIA and other leading AI labs to advance open world models. The first project from the coalition will be a base model co-developed by Runway and NVIDIA.

### Alibaba — Qwen3.7-Plus: Multimodal Agent Intelligence

- Company: Alibaba (alibaba.com)
- Announced: 2026-06-01T02:00:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://qwen.ai/blog?id=qwen3.7-plus
- Record: https://forck.live/items/4609-qwen3-7-plus-multimodal-agent-intelligence
- Subject: Qwen

Qwen3.7-Plus is a multimodal agent model that unifies vision and language, capable of operating as a multimodal interactive hybrid agent, coding agent, productivity assistant, and visual agent, with cross-harness generalization across agent frameworks. It delivers improvements in vision-language capabilities while retaining agentic strength in coding, tool use, and productivity workflows. The model is available via API on Alibaba Cloud Model Studio.

### H Company — Holo3.1 makes computer use fast and local

- Company: H Company (hcompany.ai)
- Announced: 2026-06-01
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://hcompany.ai/newsroom/holo3.1
- Record: https://forck.live/items/18369-holo3-1-makes-computer-use-fast-and-local
- Subject: Holo

H Company released the Holo3.1 family of computer-use models, improving robustness across web, desktop, and mobile environments, agent frameworks, and deployment targets. The release includes quantized checkpoints (FP8, Q4 GGUF, NVFP4) for local inference and new model sizes (0.8B, 4B, 9B) alongside the 35B-A3B model. Holo3.1 achieves gains on mobile benchmarks and introduces native function-calling protocol support.

### Factory AI — Factory Router

- Company: Factory AI (factory.com)
- Announced: 2026-06-01
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://factory.com/news/factory-router
- Record: https://forck.live/items/17846-factory-router
- Subject: Droid

Factory AI announced Factory Router, a system that automatically selects the optimal model for each Droid session to reduce token spend by 20-25% while maintaining frontier performance. It routes across providers for reliability and includes enterprise controls for model policy and routing rules. Factory Router is available in private research preview in the Factory CLI and Desktop App.

### Luma — Luma Announces The Open Physical AI Lab: An Open Science Effort to Solve Generalization in Physical AI

- Company: Luma (lumalabs.ai)
- Announced: 2026-06-01
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lumalabs.ai/news/luma-open-physical-ai-lab
- Record: https://forck.live/items/7964-luma-announces-the-open-physical-ai-lab-an-open-science-effort-to-solve
- Subject: Ray / Dream Machine

Luma announced the Open Physical AI Lab, an open science effort to solve generalization in physical AI. The lab will research and scale methods for world models and collaborate with academic and industry partners.

### xAI — Composer 2.5

- Company: xAI (x.ai)
- Announced: 2026-06-01
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/composer-2-5
- Record: https://forck.live/items/3460-composer-2-5
- Subject: Grok

Composer 2.5 is now available in Grok Build.

### MiniMax — MiniMax M3: Frontier Coding, 1M Context, Native Multimodality — All in One Model

- Company: MiniMax (minimax.io)
- Announced: 2026-06-01
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.minimax.io/blog/minimax-m3
- Record: https://forck.live/items/3396-minimax-m3-frontier-coding-1m-context-native-multimodality-all-in-one-model
- Subject: MiniMax models

MiniMax announces M3, a new model that reaches frontier capability on coding and agentic tasks, introduces MiniMax Sparse Attention (MSA) supporting up to 1M context, and is natively multimodal. It is described as the only domestic model combining all three frontier essentials and will be the only open-source one in this class.

### Cohere — RWS and Cohere build top-performing AI language intelligence for the enterprise

- Company: Cohere (cohere.com)
- Announced: 2026-06-01
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cohere.com/blog/rws-and-cohere-build-ai-language-intelligence
- Record: https://forck.live/items/3281-rws-and-cohere-build-top-performing-ai-language-intelligence-for-the-enterprise
- Subject: Command

RWS and Cohere collaborate to build top-performing AI language intelligence for the enterprise.

### OpenAI — “Data Center Bandwagon” Campaign: US-targeted influence activity

- Company: OpenAI (openai.com)
- Announced: 2026-06-01
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/disrupting-malicious-uses-of-ai-data-center-bandwagon
- Record: https://forck.live/items/99-data-center-bandwagon-campaign-us-targeted-influence-activity
- Subject: GPT / ChatGPT / API

OpenAI banned a likely PRC-origin cluster that used AI to generate social media content criticizing US data centers and AI infrastructure.

### OpenAI — “Tech and Tariffs” Campaign: Influence activity targeting US tech policy

- Company: OpenAI (openai.com)
- Announced: 2026-06-01
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/disrupting-malicious-uses-of-ai-tech-and-tariffs
- Record: https://forck.live/items/98-tech-and-tariffs-campaign-influence-activity-targeting-us-tech-policy
- Subject: GPT / ChatGPT / API

OpenAI banned accounts suspected of originating from the People's Republic of China that used AI to generate comments and cartoons about US tech policy, tariffs, and trade restrictions.