# forck.live — 23 February – 1 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-W09, 23 February – 1 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-W09 - The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md - Catalogue: https://forck.live/llms.txt ## Announcements ### Cognition — An Early Preview of SWE-1.6 and Research Update - Company: Cognition (cognition.com) - Announced: 2026-03-01 - Category: model-update - Coverage: not counted - Announcement: yes - Group: models - Source: https://cognition.com/blog/swe-1-6-preview - Record: https://forck.live/items/7676-an-early-preview-of-swe-1-6-and-research-update - Subject: Devin / SWE Cognition shared an early preview of its SWE-1.6 model, which achieves an 11% higher score than SWE-1.5 on SWE-Bench Pro. The company is rolling out early access to a small subset of users to collect feedback for tuning model behavior. The training stack has been improved to run 6x faster than three months ago. ### Vercel — How Avalara turns pipe dreams into patent-pending with v0 - Company: Vercel (vercel.com) - Announced: 2026-02-28T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/How-avalara-turns-pipedreams-into-patent-pending-with-v0 - Record: https://forck.live/items/20741-how-avalara-turns-pipe-dreams-into-patent-pending-with-v0 - Subject: v0 / AI SDK / AI Gateway Avalara used Vercel's v0 tool to create two patent-pending products in about 60 days, shifting from slide-based concepts to working prototypes built from plain-language descriptions. The company's product and design teams now start directly in v0, reducing handoffs and accelerating iteration. ### Vercel — Gamma builds design-first agents with Vercel - Company: Vercel (vercel.com) - Announced: 2026-02-28T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/gamma-builds-design-first-agents-with-vercel - Record: https://forck.live/items/20740-gamma-builds-design-first-agents-with-vercel - Subject: v0 / AI SDK / AI Gateway Gamma, a presentation platform, uses Vercel's AI SDK to build and evolve agent architectures for its Gamma Agent conversational editing feature, which launched in October 2025. The company's image pipeline has generated over 1.5 billion images across 60 models and 20 providers, and adding a new image model requires about 30 lines of code. Gamma averages more than 250 deployments per day with a 99 percent success rate, relying on Vercel's preview deployments and serverless functions. ### OpenAI — Our agreement with the Department of War - Company: OpenAI (openai.com) - Announced: 2026-02-28T12:30:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/our-agreement-with-the-department-of-war - Record: https://forck.live/items/254-our-agreement-with-the-department-of-war - Subject: GPT / ChatGPT / API OpenAI announces a contract with the Department of War, detailing safety red lines, legal protections, and deployment of AI systems in classified environments. ### Vercel — Keeping community human while scaling with agents - Company: Vercel (vercel.com) - Announced: 2026-02-27T13:00:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://vercel.com/blog/keeping-community-human-while-scaling-with-agents - Record: https://forck.live/items/20742-keeping-community-human-while-scaling-with-agents - Subject: v0 / AI SDK / AI Gateway Vercel built two agents, Community Guardian and c0, to automate community management tasks such as routing, triage, follow-ups, and research. In the first 23 days, the system handled 4,716 initial responses, revived 1 in 8 ghosted threads, and ran over 1,400 agent operations. The agents were built by a non-engineer using Vercel's AI Gateway, AI SDK, and Workflows. ### Runway — Accelerating Robot Policy Evaluation with General World Models - Company: Runway (runway.com) - Announced: 2026-02-27T10:10:10.110000+00:00 - Category: research-paper - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://runway.com/research/accelerating-robot-policy-evaluation - Record: https://forck.live/items/7895-accelerating-robot-policy-evaluation-with-general-world-models - Subject: Gen / Aleph Runway tested GWM-Robotics, a world model for robot policy evaluation, across eight manipulation policies and found a 0.95 correlation between simulated and real-world scores. The model outperforms PolaRiS and can generate up to 30-second rollouts. Runway is building GWM-Robotics with partners including NVIDIA and Berkshire Grey. ### OpenAI — OpenAI and Amazon announce strategic partnership - Company: OpenAI (openai.com) - Announced: 2026-02-27T05:30:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/amazon-partnership - Record: https://forck.live/items/258-openai-and-amazon-announce-strategic-partnership - Subject: GPT / ChatGPT / API OpenAI and Amazon announced a strategic partnership to bring OpenAI's Frontier platform to AWS, expanding AI infrastructure, custom models, and enterprise AI agents. ### OpenAI — Joint Statement from OpenAI and Microsoft - Company: OpenAI (openai.com) - Announced: 2026-02-27T05:30:00+00:00 - Category: partnership-acquisition - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/continuing-microsoft-partnership - Record: https://forck.live/items/257-joint-statement-from-openai-and-microsoft - Subject: GPT / ChatGPT / API OpenAI and Microsoft continue to collaborate closely across research, engineering, and product development, building on years of deep collaboration and shared success. ### OpenAI — Scaling AI for everyone - Company: OpenAI (openai.com) - Announced: 2026-02-27T05:30:00+00:00 - Category: funding-company-news - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/scaling-ai-for-everyone - Record: https://forck.live/items/256-scaling-ai-for-everyone - Subject: GPT / ChatGPT / API OpenAI announces $110 billion in new investment at a $730 billion pre-money valuation, with contributions from SoftBank, NVIDIA, and Amazon. ### OpenAI — Introducing the Stateful Runtime Environment for Agents in Amazon Bedrock - Company: OpenAI (openai.com) - Announced: 2026-02-27T05:30:00+00:00 - Category: capability-change - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/introducing-the-stateful-runtime-environment-for-agents-in-amazon-bedrock - Record: https://forck.live/items/255-introducing-the-stateful-runtime-environment-for-agents-in-amazon-bedrock - Subject: GPT / ChatGPT / API OpenAI announces a stateful runtime environment for agents in Amazon Bedrock, providing persistent orchestration, memory, and secure execution for multi-step AI workflows. ### Odyssey — On the Origin of Species (of World Models) - Company: Odyssey (odyssey.systems) - Announced: 2026-02-27 - Category: not stated - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://odyssey.systems/on-the-origin-of-species-of-world-models - Record: https://forck.live/items/18599-on-the-origin-of-species-of-world-models - Subject: Odyssey world models World models are entering a renaissance, but the term is being overloaded. Frequently, and badly. ### LG AI Research — [LG Accountability Report on AI Ethics EP.1] A New Benchmark for Trust at the India AI Summit - LG AI Research’s Vision for AI in Service of Humanity - Company: LG AI Research (lgresearch.ai) - Announced: 2026-02-27 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://www.lgresearch.ai/news/view?seq=627 - Record: https://forck.live/items/10995-lg-accountability-report-on-ai-ethics-ep-1-a-new-benchmark-for-trust-at-the - Subject: EXAONE LG AI Research unveiled K-AUT (Korea-Augmented Universal Taxonomy), an independently developed universal AI risk classification framework comprising 226 detailed risk categories across four domains: universal human values, social safety, Korean contextual specificity, and emerging risks. The institute also announced a joint AI Ethics MOOC with UNESCO scheduled for global launch in May, and released its 2025 AI Ethics Accountability Report documenting its technological safeguards and responsible AI practices. ### Cognition — How Cognition Uses Devin to Build Devin - Company: Cognition (cognition.com) - Announced: 2026-02-27 - Category: capability-change - Coverage: not counted - Announcement: no - Group: routine - Source: https://cognition.com/blog/how-cognition-uses-devin-to-build-devin - Record: https://forck.live/items/7677-how-cognition-uses-devin-to-build-devin - Subject: Devin / SWE Cognition describes how it uses its own product, Devin, to build and improve Devin, including automated code review, bug triage, and design system audits. ### LG AI Research — [LG Accountability Report on AI Ethics EP.1] A New Benchmark for Trust at the India AI Summit - LG AI Research’s Vision for AI in Service of Humanity - Company: LG AI Research (lgresearch.ai) - Announced: 2026-02-27 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://www.lgresearch.ai/blog/view?seq=627 - Record: https://forck.live/items/4759-lg-accountability-report-on-ai-ethics-ep-1-a-new-benchmark-for-trust-at-the - Subject: EXAONE LG AI Research participated in the India AI Summit, introduced the K-AUT risk classification framework, announced a joint AI Ethics MOOC with UNESCO for global launch in May, and released the 2025 AI Ethics Accountability Report. ### OpenAI — An update on our mental health-related work - Company: OpenAI (openai.com) - Announced: 2026-02-27 - Category: safety-policy-update - Coverage: not counted - Announcement: yes - Group: announcements - Source: https://openai.com/index/update-on-mental-health-related-work - Record: https://forck.live/items/259-an-update-on-our-mental-health-related-work - Subject: GPT / ChatGPT / API OpenAI shares updates on its mental health safety work, including parental controls, trusted contacts, improved distress detection, and recent litigation developments. ### Replit — We Built a Video Rendering Engine by Lying to the Browser About What Time It Is - Company: Replit (replit.com) - Announced: 2026-02-26T16:30:00+00:00 - Category: not stated - Coverage: not counted - Announcement: no - Group: routine - Source: https://replit.com/blog/browsers-dont-want-to-be-cameras - Record: https://forck.live/items/17918-we-built-a-video-rendering-engine-by-lying-to-the-browser-about-what-time-it-is - Subject: Replit Agent Replit built a video rendering engine that injects a virtual clock into the browser, replacing timing APIs such as setTimeout, requestAnimationFrame, and Date.now() to make animations deterministic. The engine also includes workarounds for headless Chrome's compositor and a multi-layer solution for handling