# forck.live — 9–15 March 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-W11, 9–15 March 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-W11
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Z.ai — GLM-5-Turbo

- Company: Z.ai (z.ai)
- Announced: 2026-03-15
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://docs.z.ai/release-notes/new-released#2026-03-15
- Record: https://forck.live/items/4450-glm-5-turbo
- Subject: GLM

Z.ai announces GLM-5-Turbo, a model update focused on improving stability and efficiency in long-chain agent tasks, with enhanced tool integration and instruction decomposition.

### Google Research — Protecting cities with AI-driven flash flood forecasting

- Company: Google Research (research.google)
- Announced: 2026-03-12T13:03:15+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/protecting-cities-with-ai-driven-flash-flood-forecasting/
- Record: https://forck.live/items/1359-protecting-cities-with-ai-driven-flash-flood-forecasting
- Subject: Research

Google Research announces the roll-out of Urban Flash Flood forecasts on Flood Hub, using a new AI-powered methodology that provides up to 24 hours advance notice for flash floods in urban areas. The system leverages a novel training method called Groundsource, which uses Gemini to extract ground truth from news reports, and relies on global weather products and forecasts.

### Vercel — How Notion Workers run untrusted code at scale with Vercel Sandbox

- Company: Vercel (vercel.com)
- Announced: 2026-03-12T13:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/notion-workers-vercel-sandbox
- Record: https://forck.live/items/20734-how-notion-workers-run-untrusted-code-at-scale-with-vercel-sandbox
- Subject: v0 / AI SDK / AI Gateway

Notion Workers let you write and deploy code to give Custom Agents new powers: sync external data, trigger automations, call any API.

### Harvey — Harvey in Practice: How IP Teams Manage the Patent Lifecycle

- Company: Harvey (harvey.ai)
- Announced: 2026-03-12T13:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/harvey-in-practice-how-intellectual-property-teams-use-harvey
- Record: https://forck.live/items/17596-harvey-in-practice-how-ip-teams-manage-the-patent-lifecycle
- Subject: Harvey

Harvey describes how its platform supports intellectual property teams across the patent lifecycle, including invention disclosure analysis, prior art mapping, filing document preparation, office action response drafting, post-grant brief preparation, and portfolio analysis. The post includes a testimonial from Estrella of counsel Jeffrey Hernandez-Escobar stating that Harvey reduces the time needed to analyze Office Actions and synthesize case law.

### Google Research — Introducing Groundsource: Turning news reports into data with Gemini

- Company: Google Research (research.google)
- Announced: 2026-03-12T10:42:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/introducing-groundsource-turning-news-reports-into-data-with-gemini/
- Record: https://forck.live/items/1360-introducing-groundsource-turning-news-reports-into-data-with-gemini
- Subject: Research

Google Research introduces Groundsource, a methodology that uses the Gemini LLM to extract structured historical disaster data from news reports, starting with a publicly available dataset of 2.6 million flash flood events across over 150 countries.

### Runway — Building Interactive AI Characters Responsibly

- Company: Runway (runway.com)
- Announced: 2026-03-12T10:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://runway.com/news/safety/building-interactive-ai-characters-responsibly
- Record: https://forck.live/items/7892-building-interactive-ai-characters-responsibly
- Subject: Gen / Aleph

Runway launched Runway Characters, a real-time video agent API for building custom conversational characters from a single image. The API is built on the company's World Model, GWM-1. Runway also detailed safety restrictions including filters on avatars featuring children, public figures, or IP-protected content.

### Genspark — Genspark AI Workspace 3.0: Introducing Your First AI Employee

- Company: Genspark (genspark.ai)
- Announced: 2026-03-12
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.genspark.ai/blog/genspark-ai-workspace-3
- Record: https://forck.live/items/18260-genspark-ai-workspace-3-0-introducing-your-first-ai-employee
- Subject: Genspark

### Harvey — Harvey to Open Singapore Office in June

- Company: Harvey (harvey.ai)
- Announced: 2026-03-12
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/harvey-to-open-singapore-office-in-june
- Record: https://forck.live/items/17597-harvey-to-open-singapore-office-in-june
- Subject: Harvey

Harvey announced it will open an office in Singapore in June, citing accelerating growth across APAC and deepening partnerships with legal teams in the region. The office will include legal engineering, customer success, sales, and G&A roles.

### Microsoft AI — Introducing Copilot Health

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-03-12
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://microsoft.ai/news/introducing-copilot-health
- Record: https://forck.live/items/8433-introducing-copilot-health
- Subject: MAI

Microsoft launched Copilot Health, a secure space within Copilot that integrates health records, wearable data, and health history to provide personalized health insights. It is rolling out first in English in the United States to adults aged 18 and older, with a waitlist for early access. The feature uses AI to analyze patterns, includes citations from credible health organizations, and has achieved ISO/IEC 42001 certification.

### Together AI — Build real-time voice agents on Together AI

- Company: Together AI (together.ai)
- Announced: 2026-03-12
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.together.ai/blog/build-real-time-voice-agents-on-together-ai
- Record: https://forck.live/items/2370-build-real-time-voice-agents-on-together-ai
- Subject: Inference platform

Together AI launched a unified solution for building real-time voice agents, featuring co-located STT, LLM, and TTS infrastructure, native integrations with Cartesia and Deepgram models, and capabilities for low-latency, production-scale deployments.

### Google Research — Exploring the feasibility of conversational diagnostic AI in a real-world clinical study

- Company: Google Research (research.google)
- Announced: 2026-03-11T16:58:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/exploring-the-feasibility-of-conversational-diagnostic-ai-in-a-real-world-clinical-study/
- Record: https://forck.live/items/1361-exploring-the-feasibility-of-conversational-diagnostic-ai-in-a-real-world
- Subject: Research

Google Research and Google DeepMind, in partnership with Beth Israel Deaconess Medical Center, conducted a prospective single-center feasibility study of AMIE, a conversational medical AI for clinical reasoning and dialogue. The study involved 100 adult patients in a real-world ambulatory primary care setting, assessing safety, feasibility, clinical reasoning, and patient/physician perceptions. Results showed zero safety interventions, increased patient trust, and diagnostic performance on par with primary care physicians.

### Harvey — Introducing BigLaw Bench: Research

- Company: Harvey (harvey.ai)
- Announced: 2026-03-11T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/introducing-big-law-bench-research
- Record: https://forck.live/items/17598-introducing-biglaw-bench-research
- Subject: Harvey

Harvey introduced BigLaw Bench: Research, a dataset of hard US Case Law research problems that current models cannot solve even with search tools. The benchmark requires models to use tools for searches and provide grounded, cited responses, aiming to identify model improvements and failure modes in legal research. Harvey will use the benchmark to evaluate models' research capabilities across practice areas such as corporate, securities litigation, and privacy.

### Replit — The Future is Actually Very Human

- Company: Replit (replit.com)
- Announced: 2026-03-11T13:40:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/replit-raises-400-million-dollars
- Record: https://forck.live/items/17915-the-future-is-actually-very-human
- Subject: Replit Agent

Replit announced a $400 million funding round at a $9 billion valuation, a 3x increase in six months. The company also introduced Agent 4, described as 10x faster than Agent 3, designed to handle background tasks while users focus on creative work. Investors include Georgian, G Squared, Prysm Capital, 1789 Capital, YC, Coatue, a16z, Craft Ventures, and Qatar Investment Authority, among others.

### Replit — Introducing Replit Agent 4: Built for Creativity

- Company: Replit (replit.com)
- Announced: 2026-03-11T13:30:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/introducing-agent-4-built-for-creativity
- Record: https://forck.live/items/17916-introducing-replit-agent-4-built-for-creativity
- Subject: Replit Agent

Introducing Agent 4 — our fastest, most versatile Agent yet. It's built around a simple idea: you should spend your time creating, not coordinating.

### Lovable — 9 ways to use Workspace Knowledge to empower your entire team

- Company: Lovable (lovable.dev)
- Announced: 2026-03-11T13:17:31+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/9-ways-to-use-workspace-knowledge-to-empower-your-entire-team
- Record: https://forck.live/items/12097-9-ways-to-use-workspace-knowledge-to-empower-your-entire-team
- Subject: Lovable / AI app builder

Lovable's Workspace Knowledge feature allows workspace admins to define persistent instructions—such as coding standards, testing requirements, library preferences, and architectural guardrails—that are automatically applied to every project in the workspace. This enables non-technical team members to build with consistent code and quality practices without needing to understand the underlying rules.

### Runway — Introducing Runway Labs

- Company: Runway (runway.com)
- Announced: 2026-03-11T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://runway.com/news/introducing-runway-labs
- Record: https://forck.live/items/7893-introducing-runway-labs
- Subject: Gen / Aleph

Runway announced Runway Labs, an internal incubator to explore new applications for generative video and General World Models, led by co-founder Alejandro Matamala Ortiz. The incubator will prototype experimental products and they are hiring for several engineering roles.

### OpenAI — Designing AI agents to resist prompt injection

- Company: OpenAI (openai.com)
- Announced: 2026-03-11T11:30:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/designing-agents-to-resist-prompt-injection
- Record: https://forck.live/items/231-designing-ai-agents-to-resist-prompt-injection
- Subject: GPT / ChatGPT / API

OpenAI describes how ChatGPT defends against prompt injection and social engineering by constraining risky actions and protecting sensitive data in agent workflows.

### OpenAI — From model to agent: Equipping the Responses API with a computer environment

- Company: OpenAI (openai.com)
- Announced: 2026-03-11T11:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/equip-responses-api-computer-environment
- Record: https://forck.live/items/232-from-model-to-agent-equipping-the-responses-api-with-a-computer-environment
- Subject: GPT / ChatGPT / API

OpenAI equipped the Responses API with an agent runtime that uses a shell tool and hosted containers to run secure, scalable agents with files, tools, and state.

### Factory AI — Factory Analytics

- Company: Factory AI (factory.com)
- Announced: 2026-03-11
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://factory.com/news/factory-analytics
- Record: https://forck.live/items/17837-factory-analytics
- Subject: Droid

Factory introduced Factory Analytics, a dashboard that tracks Droid usage, token consumption, tool calls, activity, productivity, and per-user metrics. The feature is available for Enterprise plan customers and includes API access, OpenTelemetry export, and AI-generated reports.

### Cartesia — Mamba-3: An Inference-First State Space Model

- Company: Cartesia (cartesia.ai)
- Announced: 2026-03-11
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.cartesia.ai/blog/mamba-3
- Record: https://forck.live/items/8012-mamba-3-an-inference-first-state-space-model
- Subject: Sonic / Ink

Cartesia and Goomba Lab introduced Mamba-3, a state space model designed with inference efficiency as the primary goal. The model improves upon Mamba-2 through a more expressive recurrence, complex-valued SSM systems, and multi-input multi-output SSMs, aiming to push the quality-efficiency frontier while maintaining similar inference latency.

### Mistral AI — Rails testing on autopilot: Building an agent that writes what developers won't

- Company: Mistral AI (mistral.ai)
- Announced: 2026-03-11
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://mistral.ai/news/rails-testing-on-autopilot-building-an-agent-that-writes-what-developers-wont
- Record: https://forck.live/items/3303-rails-testing-on-autopilot-building-an-agent-that-writes-what-developers-won-t
- Subject: Mistral / Le Chat

Mistral AI's Applied AI Proto team is building an agent to automate Rails testing.

### Cursor — New Plugins on the Cursor Marketplace

- Company: Cursor (cursor.com)
- Announced: 2026-03-11
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/03-11-26
- Record: https://forck.live/items/2626-new-plugins-on-the-cursor-marketplace
- Subject: Cursor

Cursor announced the addition of over 30 new plugins from various partners, enabling the platform to read, write, and take actions across more of the user's stack.

### Together AI — Together AI Brings NVIDIA Nemotron 3 to Developers on Day 0

- Company: Together AI (together.ai)
- Announced: 2026-03-11
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.together.ai/blog/nvidia-nemotron-3-super
- Record: https://forck.live/items/2371-together-ai-brings-nvidia-nemotron-3-to-developers-on-day-0
- Subject: Inference platform

Together AI announced the availability of NVIDIA Nemotron 3 Super, a 120B-parameter hybrid MoE model with a 1M-token context window, on its Dedicated Inference platform.

### OpenAI — Wayfair boosts catalog accuracy and support speed with OpenAI

- Company: OpenAI (openai.com)
- Announced: 2026-03-11
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/wayfair
- Record: https://forck.live/items/233-wayfair-boosts-catalog-accuracy-and-support-speed-with-openai
- Subject: GPT / ChatGPT / API

OpenAI published a customer story about Wayfair using OpenAI models to improve ecommerce support and product catalog accuracy by automating ticket triage and enhancing product attributes at scale.

### Google DeepMind — 10 years of AlphaGo: The turning point for AI

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-03-10T17:28:31+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.youtube.com/watch?v=qoinGjj60Fo
- Record: https://forck.live/items/19944-10-years-of-alphago-the-turning-point-for-ai

Google DeepMind published a retrospective video marking the 10-year anniversary of AlphaGo's 2016 match against Lee Sedol, which it describes as a turning point that sparked the modern AI revolution and laid foundations for later achievements such as protein folding.

### Harvey — The Brief: March 2026

- Company: Harvey (harvey.ai)
- Announced: 2026-03-10T15:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/the-brief-march-2026
- Record: https://forck.live/items/17599-the-brief-march-2026
- Subject: Harvey

Harvey added secure collaboration features including shared workspaces, a mobile app, and workflow suggestions. It also expanded analysis capabilities with Anthropic's Opus 4.6 and Sonnet 4.6 models, increased file upload limits to 500 MB, and added review tables and deep analysis in custom agents. Legal research was broadened with a FromCounsel integration and over 110 new regional data sources.

### Vercel — How we run Vercel's CDN in front of Discourse

- Company: Vercel (vercel.com)
- Announced: 2026-03-10T13:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/how-we-run-vercels-cdn-in-front-of-discourse
- Record: https://forck.live/items/20735-how-we-run-vercel-s-cdn-in-front-of-discourse
- Subject: v0 / AI SDK / AI Gateway

Vercel published a guide explaining how it uses its own CDN to proxy the Vercel Community forum, which runs on Discourse, adding firewall protection, DDoS mitigation, and observability without migrating the application. The setup uses two domains—an inner host for the origin server and an outer host for the Vercel project—and rewrites traffic via vercel.ts. Vercel also uses microfrontends to mount a Next.js app on the same domain for pages that are impractical to implement as Discourse plugins.

### Harvey — Why Harvey is Multi-Model by Design

- Company: Harvey (harvey.ai)
- Announced: 2026-03-10T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/why-harvey-is-multi-model-by-design
- Record: https://forck.live/items/17600-why-harvey-is-multi-model-by-design
- Subject: Harvey

Harvey describes its multi-model platform, which integrates models from Anthropic, Google DeepMind, and OpenAI to maximize performance, ensure continuity, and give customers control over model selection. The architecture provides structural redundancy so that if one provider experiences an outage, work can be routed to an alternative model without disrupting the user's workflow. Harvey also enforces consistent security standards across all providers, including zero data retention and no training on customer inputs.

### OpenAI — Improving instruction hierarchy in frontier LLMs

- Company: OpenAI (openai.com)
- Announced: 2026-03-10T11:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/instruction-hierarchy-challenge
- Record: https://forck.live/items/235-improving-instruction-hierarchy-in-frontier-llms
- Subject: GPT / ChatGPT / API

OpenAI introduces the IH-Challenge method to train models to better prioritize trusted instructions, enhancing instruction hierarchy, safety steerability, and resistance to prompt injection attacks.

### OpenAI — New ways to learn math and science in ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-03-10T10:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/new-ways-to-learn-math-and-science-in-chatgpt
- Record: https://forck.live/items/236-new-ways-to-learn-math-and-science-in-chatgpt
- Subject: GPT / ChatGPT / API

ChatGPT introduces interactive visual explanations for math and science, helping students explore formulas, variables, and concepts in real time.

### Thinking Machines Lab — Thinking Machines Lab and NVIDIA Announce Long-Term Gigawatt-Scale Strategic Partnership

- Company: Thinking Machines Lab (thinkingmachines.ai)
- Announced: 2026-03-10
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://thinkingmachines.ai/news/nvidia-partnership
- Record: https://forck.live/items/18560-thinking-machines-lab-and-nvidia-announce-long-term-gigawatt-scale-strategic
- Subject: Thinking Machines / Inkling

Thinking Machines Lab and NVIDIA announced a multi-year strategic partnership to deploy at least one gigawatt of next-generation NVIDIA Vera Rubin systems for frontier model training and customizable AI platforms. NVIDIA also made a significant investment in Thinking Machines Lab to support long-term growth.

### Microsoft AI — Health Check: How People Use Copilot for Health

- Company: Microsoft AI (microsoft.ai)
- Announced: 2026-03-10
- Category: research-paper
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://microsoft.ai/news/health-check-how-people-use-copilot-for-health
- Record: https://forck.live/items/8434-health-check-how-people-use-copilot-for-health
- Subject: MAI

Microsoft AI analyzed over half a million health-related conversations with Copilot from January 2026, finding that people primarily ask about symptoms (10.9%), lifestyle coaching (9%), and healthcare navigation (5.8%). Nearly one in five conversations involve personal symptom assessment or condition management, and one in seven are on behalf of someone else. Mobile usage skews toward personal health topics, while desktop is used for research.

### Together AI — New in Together GPU Clusters: Autoscaling, observability, and self-healing

- Company: Together AI (together.ai)
- Announced: 2026-03-10
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.together.ai/blog/new-in-together-gpu-clusters-autoscaling-observability-self-healing
- Record: https://forck.live/items/2372-new-in-together-gpu-clusters-autoscaling-observability-and-self-healing
- Subject: Inference platform

Together AI announces major enterprise enhancements to Together GPU Clusters, including autoscaling, RBAC, full-stack observability, self-serve node repair, and active health checks.

### Hugging Face — Keep the Tokens Flowing: Lessons from 16 Open-Source RL Libraries

- Company: Hugging Face (huggingface.co)
- Announced: 2026-03-10
- Category: research-paper
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/async-rl-training-landscape
- Record: https://forck.live/items/1536-keep-the-tokens-flowing-lessons-from-16-open-source-rl-libraries
- Subject: Platform

Hugging Face researchers surveyed 16 open-source libraries implementing async RL training patterns, comparing them across seven design axes, and distill lessons for building async trainers.

### Hugging Face — Introducing Storage Buckets on the Hugging Face Hub

- Company: Hugging Face (huggingface.co)
- Announced: 2026-03-10
- Category: infrastructure-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/storage-buckets
- Record: https://forck.live/items/1535-introducing-storage-buckets-on-the-hugging-face-hub
- Subject: Platform

Hugging Face introduces Storage Buckets, a mutable, S3-like object storage on the Hub for ML intermediate files, backed by Xet for chunk-based deduplication, with CLI, Python, and JavaScript support, and pre-warming for data locality.

### Google DeepMind — From games to biology and beyond: 10 years of AlphaGo’s impact

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-03-09T13:52:36+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://deepmind.google/blog/10-years-of-alphago/
- Record: https://forck.live/items/1248-from-games-to-biology-and-beyond-10-years-of-alphago-s-impact
- Subject: Gemini

The blog post reflects on the ten-year impact of AlphaGo, discussing how it has catalyzed scientific discovery and paved a path toward artificial general intelligence (AGI).

### Harvey — Introducing Agent Builder: Build Smarter Agents for Complex Legal Work

- Company: Harvey (harvey.ai)
- Announced: 2026-03-09T11:55:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/introducing-agent-builder
- Record: https://forck.live/items/17601-introducing-agent-builder-build-smarter-agents-for-complex-legal-work
- Subject: Harvey

Harvey evolved its Workflow Builder into Agent Builder, enabling users to create custom Workflow agents tailored to specific teams or processes. The new agents leverage improved model reasoning to handle multi-step tasks with less detailed instructions, include human-in-the-loop checkpoints, and will support background scheduling for autonomous execution.

### Runway — Introducing Runway Characters

- Company: Runway (runway.com)
- Announced: 2026-03-09T10:00:00+00:00
- Category: api-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://runway.com/news/introducing-runway-characters
- Record: https://forck.live/items/7894-introducing-runway-characters
- Subject: Gen / Aleph

Runway launched Characters, a real-time video agent API for building custom conversational avatars from a single image with no fine-tuning, using the company's GWM-1 model. Characters is available via the developer platform for enterprise teams and as preset avatars in the Runway web app for consumers. Partners including BBC and Silverside are already using it.

### OpenAI — OpenAI to acquire Promptfoo

- Company: OpenAI (openai.com)
- Announced: 2026-03-09T10:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/openai-to-acquire-promptfoo
- Record: https://forck.live/items/237-openai-to-acquire-promptfoo
- Subject: GPT / ChatGPT / API

OpenAI announces its acquisition of Promptfoo, an AI security platform for enterprise vulnerability identification and remediation.

### Krea — Krea Edit: Region Edit, Colorize & Relight Any Image

- Company: Krea (krea.ai)
- Announced: 2026-03-09
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.krea.ai/blog/krea-edit
- Record: https://forck.live/items/15086-krea-edit-region-edit-colorize-relight-any-image
- Subject: Krea 2 / AI creative platform

Krea launched Krea Edit, a rebuilt editing tool that allows users to select specific regions of an image and describe changes, with the AI altering only that area while preserving the rest. The tool also includes features for colorizing grayscale images, relighting scenes with virtual lights, and re-rendering images through a camera lens simulator. Users can access Krea Edit from generated images or directly at krea.ai/edit, with each edit stacking and tracked in a generation history strip.

### Hugging Face — LeRobot v0.5.0: Scaling Every Dimension

- Company: Hugging Face (huggingface.co)
- Announced: 2026-03-09
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/lerobot-release-v050
- Record: https://forck.live/items/1538-lerobot-v0-5-0-scaling-every-dimension
- Subject: Platform

Hugging Face announced LeRobot v0.5.0, a major update to its open-source robotics library that adds new hardware support (including the Unitree G1 humanoid), six new policies (e.g., Pi0-FAST, Real-Time Chunking), faster dataset streaming and encoding, Hub-browseable environments (EnvHub), NVIDIA IsaacLab-Arena integration, and a codebase upgrade to Python 3.12+ and Transformers v5.

### Hugging Face — Ulysses Sequence Parallelism: Training with Million-Token Contexts

- Company: Hugging Face (huggingface.co)
- Announced: 2026-03-09
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/ulysses-sp
- Record: https://forck.live/items/1537-ulysses-sequence-parallelism-training-with-million-token-contexts
- Subject: Platform

Ulysses Sequence Parallelism, a technique for training large language models on long sequences by distributing attention computation across multiple GPUs via attention head parallelism. It covers the methodology, communication complexity, and integration with Hugging Face's Accelerate library.