# forck.live — 4–10 May 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-W19, 4–10 May 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-W19
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### Vercel — How Superset built the IDE for AI agents on Vercel

- Company: Vercel (vercel.com)
- Announced: 2026-05-10T14:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/how-superset-built-the-ide-for-ai-agents-on-vercel
- Record: https://forck.live/items/20710-how-superset-built-the-ide-for-ai-agents-on-vercel
- Subject: v0 / AI SDK / AI Gateway

Vercel published a case study about Superset, an IDE for multi-agent development that runs up to 10 coding agents in parallel. Superset uses Vercel's platform for preview deployments, AI SDK, AI Gateway, Blob storage, Fluid compute, Cron Jobs, and BotID. The company reports 1,000–1,400 deployments per week, ~600 preview deployments per day, ~30 second average build time, and 57–64% week-over-week DAU growth.

### Genspark — Seeing AGI (8): The Rebirth of the Human Role

- Company: Genspark (genspark.ai)
- Announced: 2026-05-10
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.genspark.ai/blog/seeing-agi-8-rebirth-human-role
- Record: https://forck.live/items/18256-seeing-agi-8-the-rebirth-of-the-human-role
- Subject: Genspark

### Baidu — ERNIE 5.1 Officially Released! Topping Multiple Leaderboards — A Model That Writes Better and Understands You More

- Company: Baidu (baidu.com)
- Announced: 2026-05-09
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://ernie.baidu.com/blog/posts/ernie-5.1-0508-release/
- Record: https://forck.live/items/4640-ernie-5-1-officially-released-topping-multiple-leaderboards-a-model-that
- Subject: ERNIE

Baidu officially released ERNIE 5.1, a new model derived from ERNIE 5.0 with compressed parameters (total to ~1/3, active to ~1/2) and pre-training cost at only 6% of comparable models. It uses a disaggregated fully-asynchronous reinforcement learning infrastructure. ERNIE 5.1 achieved 4th place globally and 1st among Chinese models on the Arena Search leaderboard, and shows strong agentic, reasoning, and knowledge capabilities. A playground is available on Baidu AI Studio.

### Sakana AI — Sparser, Faster, Lighter Transformer Language Models

- Company: Sakana AI (sakana.ai)
- Announced: 2026-05-08T15:00:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://sakana.ai/twell
- Record: https://forck.live/items/4684-sparser-faster-lighter-transformer-language-models
- Subject: Sakana models

Sakana AI, in collaboration with NVIDIA, announces a paper accepted at ICML 2026 introducing open-source GPU kernels and a new sparse packing format (TwELL) to speed up inference and training of sparse transformer language models, achieving >20% speedups and reduced memory usage.

### OpenAI — Running Codex safely at OpenAI

- Company: OpenAI (openai.com)
- Announced: 2026-05-08T12:30:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/running-codex-safely
- Record: https://forck.live/items/138-running-codex-safely-at-openai
- Subject: GPT / ChatGPT / API

OpenAI describes how it runs Codex securely using sandboxing, approvals, network policies, and agent-native telemetry.

### Runway — Our Approach to Child Safety

- Company: Runway (runway.com)
- Announced: 2026-05-08T10:00:00+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://runway.com/news/safety/our-approach-to-child-safety
- Record: https://forck.live/items/7885-our-approach-to-child-safety
- Subject: Gen / Aleph

Runway described its child safety approach across model development, deployment, and maintenance, including training data safeguards, red teaming, usage policy prohibiting CSAM, detection systems, and reporting to NCMEC.

### Supabase — Supabase Is Now an Official ChatGPT App

- Company: Supabase (supabase.com)
- Announced: 2026-05-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/supabase-is-now-an-official-chatgpt-app
- Record: https://forck.live/items/21286-supabase-is-now-an-official-chatgpt-app
- Subject: Supabase

Supabase is now an official app in ChatGPT, offering 29 tools for database management, project operations, branching, edge functions, and documentation search. Users on any Supabase plan and a paid ChatGPT plan can connect their projects and manage infrastructure through natural language prompts.

### Together AI — Deploy and inference any model from HuggingFace

- Company: Together AI (together.ai)
- Announced: 2026-05-08
- Category: developer-tool-release
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.together.ai/blog/deploy-and-inference-any-model-from-huggingface
- Record: https://forck.live/items/2350-deploy-and-inference-any-model-from-huggingface
- Subject: Inference platform

Together AI announces a new skill for Goose, a CLI agent runner, that allows developers to deploy any HuggingFace model on Together's Dedicated Container Inference (DCI) infrastructure with a single prompt. The post demonstrates deploying Netflix's void-model on release day.

### Harvey — Harvey in Practice: How In-House Teams Monitor, Interpret, and Act on Regulatory Change

- Company: Harvey (harvey.ai)
- Announced: 2026-05-07T18:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/harvey-in-practice-in-house-regulatory-and-compliance
- Record: https://forck.live/items/17569-harvey-in-practice-how-in-house-teams-monitor-interpret-and-act-on-regulatory
- Subject: Harvey

Harvey published a guide showing how in-house legal teams use its platform for regulatory compliance, including monitoring changes, analyzing applicability, mapping obligations, identifying policy gaps, and reviewing contracts. The guide cites Bridgewater, where Harvey reduced review time for complex trading agreements by over 95%, and Talanx, where Harvey is embedded across compliance workflows.

### Replit — Security Center 2.0: Act on vulnerabilities in bulk across all your apps

- Company: Replit (replit.com)
- Announced: 2026-05-07T17:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://replit.com/blog/security-center
- Record: https://forck.live/items/17903-security-center-2-0-act-on-vulnerabilities-in-bulk-across-all-your-apps
- Subject: Replit Agent

Replit introduced Security Center 2.0, which provides a centralized view of vulnerabilities across all projects in a business account and enables bulk actions such as notifying project owners or unpublishing apps. The update also adds a per-project "Fix with Agent" feature that creates a task with a proposed patch for the project owner to review and apply.

### Harvey — The Supreme Case for Harvey

- Company: Harvey (harvey.ai)
- Announced: 2026-05-07T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/the-supreme-case-for-harvey
- Record: https://forck.live/items/17570-the-supreme-case-for-harvey
- Subject: Harvey

Harvey describes how its custom tool, Harvey Moot, was used by litigators to prepare for a Supreme Court case, and announces plans to bring the tool to law school classrooms for legal education.

### Google DeepMind — Find out how AlphaEvolve has gone from research to solving real-life problems.

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-05-07T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://blog.google/innovation-and-ai/infrastructure-and-cloud/google-cloud/alphaevolve-updates/
- Record: https://forck.live/items/1314-find-out-how-alphaevolve-has-gone-from-research-to-solving-real-life-problems
- Subject: Gemini

The source text states that AlphaEvolve has become a powerful engine for scientific and societal progress.

### OpenAI — Scaling Trusted Access for Cyber with GPT-5.5 and GPT-5.5-Cyber

- Company: OpenAI (openai.com)
- Announced: 2026-05-07T13:00:00+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/gpt-5-5-with-trusted-access-for-cyber
- Record: https://forck.live/items/139-scaling-trusted-access-for-cyber-with-gpt-5-5-and-gpt-5-5-cyber
- Subject: GPT / ChatGPT / API

OpenAI introduces GPT-5.5 and GPT-5.5-Cyber to expand Trusted Access for Cyber, aiding verified defenders in vulnerability research and critical infrastructure protection.

### Lovable — Wiz findings now in Lovable

- Company: Lovable (lovable.dev)
- Announced: 2026-05-07T12:30:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lovable.dev/blog/wiz-findings-now-in-lovable
- Record: https://forck.live/items/12085-wiz-findings-now-in-lovable
- Subject: Lovable / AI app builder

Lovable announced a native integration with Wiz that brings Wiz security scanning directly into the Lovable build experience. When connected, Wiz scans run automatically as part of Lovable's security suite, and findings appear in the Security view alongside existing built-in scanners. Results also flow into Wiz's Code and Build scans page, allowing security teams to see Lovable-built applications alongside other monitored projects.

### ElevenLabs — ElevenLabs expands presence in Australia and New Zealand

- Company: ElevenLabs (elevenlabs.io)
- Announced: 2026-05-07T11:30:00+00:00
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://elevenlabs.io/blog/elevenlabs-expands-presence-in-australia-new-zealand
- Record: https://forck.live/items/3358-elevenlabs-expands-presence-in-australia-and-new-zealand
- Subject: Voice AI

ElevenLabs is expanding its presence in Australia and New Zealand.

### OpenAI — Parloa builds service agents customers want to talk to

- Company: OpenAI (openai.com)
- Announced: 2026-05-07T11:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/parloa
- Record: https://forck.live/items/140-parloa-builds-service-agents-customers-want-to-talk-to
- Subject: GPT / ChatGPT / API

Parloa uses OpenAI models to power scalable, voice-driven AI customer service agents for enterprises.

### OpenAI — Advancing voice intelligence with new models in the API

- Company: OpenAI (openai.com)
- Announced: 2026-05-07T10:00:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/advancing-voice-intelligence-with-new-models-in-the-api
- Record: https://forck.live/items/141-advancing-voice-intelligence-with-new-models-in-the-api
- Subject: GPT / ChatGPT / API

OpenAI announces new realtime voice models available in the API, capable of reasoning, translation, and transcription, aiming to enable more natural and intelligent voice experiences.

### Cursor — PR Review, Build Plan in Parallel, and Split PRs

- Company: Cursor (cursor.com)
- Announced: 2026-05-07
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/05-07-26
- Record: https://forck.live/items/2612-pr-review-build-plan-in-parallel-and-split-prs
- Subject: Cursor

Cursor 3 adds a new PR review experience, parallel build execution from plans, a quick action to split changes into PRs, and the ability to pin skills as quick-action pills. It also includes several improvements and bug fixes.

### OpenAI — Simplex rethinks software development with Codex

- Company: OpenAI (openai.com)
- Announced: 2026-05-07
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/simplex
- Record: https://forck.live/items/144-simplex-rethinks-software-development-with-codex
- Subject: GPT / ChatGPT / API

Simplex uses ChatGPT Enterprise and Codex to reduce software development time and scale AI-driven workflows.

### OpenAI — Introducing Trusted Contact in ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-05-07
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/introducing-trusted-contact-in-chatgpt
- Record: https://forck.live/items/143-introducing-trusted-contact-in-chatgpt
- Subject: GPT / ChatGPT / API

OpenAI announced a new optional safety feature in ChatGPT that notifies a trusted contact if serious self-harm concerns are detected.

### OpenAI — Testing ads in ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-05-07
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/testing-ads-in-chatgpt
- Record: https://forck.live/items/142-testing-ads-in-chatgpt
- Subject: GPT / ChatGPT / API

OpenAI is testing ads in ChatGPT to support free access, with clear labeling, answer independence, privacy protections, and user control.

### Hugging Face — vLLM V0 to V1: Correctness Before Corrections in RL

- Company: Hugging Face (huggingface.co)
- Announced: 2026-05-06T19:06:55+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/ServiceNow-AI/correctness-before-corrections
- Record: https://forck.live/items/1506-vllm-v0-to-v1-correctness-before-corrections-in-rl
- Subject: Platform

ServiceNow-AI describes migrating their RL training pipeline from vLLM V0 to V1, identifying and fixing four issues (logprob semantics, runtime defaults, inflight weight updates, fp32 lm_head) to achieve training parity.

### Harvey — How we Built Image Understanding for Legal Documents

- Company: Harvey (harvey.ai)
- Announced: 2026-05-06T18:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/building-image-understanding-for-legal-documents
- Record: https://forck.live/items/17571-how-we-built-image-understanding-for-legal-documents
- Subject: Harvey

Harvey introduced an on-demand visual analysis system for its legal AI assistant. The system renders document pages at high resolution and uses a vision model to read charts, diagrams, and complex layouts only when a user's query requires visual understanding. It first searches document text to identify candidate pages, and gates image analysis behind a text search step to avoid unnecessary processing.

### Replit — Secure more apps with External Access Tokens and Private Publishing

- Company: Replit (replit.com)
- Announced: 2026-05-06T16:06:10+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://replit.com/blog/secure-more-apps
- Record: https://forck.live/items/17904-secure-more-apps-with-external-access-tokens-and-private-publishing
- Subject: Replit Agent

At Replit, we want all users to be able to build secure applications. In addition to features like Security Agent and Auto Protect, you can restrict access to your apps with Private Publishing.

### Google DeepMind — AlphaEvolve: How our Gemini-powered coding agent is scaling impact across fields

- Company: Google DeepMind (deepmind.google)
- Announced: 2026-05-06T10:43:49+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://deepmind.google/blog/alphaevolve-impact/
- Record: https://forck.live/items/1235-alphaevolve-how-our-gemini-powered-coding-agent-is-scaling-impact-across-fields
- Subject: Gemini

AlphaEvolve's Gemini-powered algorithms are driving impact across business, infrastructure, and science.

### OpenAI — How ChatGPT learns about the world while protecting privacy

- Company: OpenAI (openai.com)
- Announced: 2026-05-06T08:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/how-chatgpt-protects-privacy
- Record: https://forck.live/items/145-how-chatgpt-learns-about-the-world-while-protecting-privacy
- Subject: GPT / ChatGPT / API

The source describes how ChatGPT protects user privacy, minimizes personal data used in training, and allows users to control whether their conversations contribute to model improvement.

### Supabase — Introducing @supabase/server

- Company: Supabase (supabase.com)
- Announced: 2026-05-06
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/introducing-supabase-server
- Record: https://forck.live/items/21287-introducing-supabase-server
- Subject: Supabase

Supabase released @supabase/server in public beta, a package that automates auth verification, client setup, request context, and common server-side boilerplate for Edge Functions, Vercel Functions, Cloudflare Workers, Hono, and Bun. It provides a SupabaseContext with pre-configured user-scoped and admin clients, declarative access control, and support for new asymmetric JWT signing keys without manual configuration.

### Perplexity — Introducing Finance Search in the Agent API

- Company: Perplexity (perplexity.ai)
- Announced: 2026-05-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.perplexity.ai/hub/blog/introducing-finance-search-in-the-agent-api
- Record: https://forck.live/items/10749-introducing-finance-search-in-the-agent-api
- Subject: Perplexity

Perplexity launched Finance Search in its Agent API, a tool that retrieves time-sensitive financial data from licensed sources in a single call. It returns cited results in a consistent schema, enabling developers to build valuation lookups, earnings recaps, and market monitors without integrating multiple providers separately. Benchmark results on FinSearchComp T1 show higher accuracy and lower cost per correct answer compared to generic web search.

### Amp — Amp, Rebuilt

- Company: Amp (ampcode.com)
- Announced: 2026-05-06
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://ampcode.com/news/neo
- Record: https://forck.live/items/7749-amp-rebuilt
- Subject: Amp

Amp released a rebuilt version of its CLI, codenamed Neo, featuring remote control, automatic context compaction, a plugin API, message queuing and steering, and a change to default permissions. The update is designed for longer-running agent tasks and compatibility with frontier models.

### xAI — Grok Imagine Quality Mode API

- Company: xAI (x.ai)
- Announced: 2026-05-06
- Category: api-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/grok-imagine-quality-mode
- Record: https://forck.live/items/3470-grok-imagine-quality-mode-api
- Subject: Grok

xAI announced the Grok Imagine Quality Mode API with higher realism, stronger text rendering, and better creative control.

### xAI — Connectors in web, iOS, and Android

- Company: xAI (x.ai)
- Announced: 2026-05-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/grok-connectors
- Record: https://forck.live/items/3469-connectors-in-web-ios-and-android
- Subject: Grok

xAI announced Connectors for Grok on web, iOS, and Android, enabling deep integrations with everyday apps.

### xAI — New Compute Partnership with Anthropic

- Company: xAI (x.ai)
- Announced: 2026-05-06
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://x.ai/news/anthropic-compute-partnership
- Record: https://forck.live/items/3468-new-compute-partnership-with-anthropic
- Subject: Grok

xAI (SpaceXAI) has signed a compute partnership with Anthropic to provide access to Colossus 1.

### Cursor — Context Usage Breakdown

- Company: Cursor (cursor.com)
- Announced: 2026-05-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/05-06-26
- Record: https://forck.live/items/2613-context-usage-breakdown
- Subject: Cursor

Cursor now provides a breakdown of the agent's context usage, allowing users to diagnose context issues and improve setup across rules, skills, MCPs, and subagents.

### Hugging Face — Adding Benchmaxxer Repellant to the Open ASR Leaderboard

- Company: Hugging Face (huggingface.co)
- Announced: 2026-05-06
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/open-asr-leaderboard-private-data
- Record: https://forck.live/items/1507-adding-benchmaxxer-repellant-to-the-open-asr-leaderboard
- Subject: Platform

Hugging Face adds private, high-quality English ASR datasets from Appen Inc. and DataoceanAI to the Open ASR Leaderboard to prevent benchmaxxing and improve evaluation trustworthiness. The leaderboard's average WER remains computed on public datasets by default, with an optional toggle to include private datasets.

### OpenAI — Uber uses OpenAI to help people earn smarter and book faster

- Company: OpenAI (openai.com)
- Announced: 2026-05-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/uber
- Record: https://forck.live/items/149-uber-uses-openai-to-help-people-earn-smarter-and-book-faster
- Subject: GPT / ChatGPT / API

Uber is using OpenAI to power AI assistants and voice features that help drivers earn smarter and riders book faster.

### OpenAI — How frontier firms are pulling ahead

- Company: OpenAI (openai.com)
- Announced: 2026-05-06
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/introducing-b2b-signals
- Record: https://forck.live/items/148-how-frontier-firms-are-pulling-ahead
- Subject: GPT / ChatGPT / API

OpenAI announces B2B Signals research, which examines how frontier enterprises are deepening AI adoption, scaling Codex-powered agentic workflows, and building durable competitive advantage.

### OpenAI — Introducing ChatGPT Futures: Class of 2026

- Company: OpenAI (openai.com)
- Announced: 2026-05-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/introducing-chatgpt-futures-class-of-2026
- Record: https://forck.live/items/147-introducing-chatgpt-futures-class-of-2026
- Subject: GPT / ChatGPT / API

OpenAI announces the ChatGPT Futures Class of 2026, a group of 26 student innovators using AI to build, research, and drive real-world impact.

### OpenAI — Singular Bank helps bankers move fast with ChatGPT and Codex

- Company: OpenAI (openai.com)
- Announced: 2026-05-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/singular-bank
- Record: https://forck.live/items/146-singular-bank-helps-bankers-move-fast-with-chatgpt-and-codex
- Subject: GPT / ChatGPT / API

Singular Bank built an internal assistant called Singularity, using ChatGPT and Codex, which helps bankers save 60–90 minutes daily on meeting preparation, portfolio analysis, and follow-up tasks.

### Harvey — Legal Agents for Every Matter, Tailored to You

- Company: Harvey (harvey.ai)
- Announced: 2026-05-05T12:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/agentic-platform-updates
- Record: https://forck.live/items/17572-legal-agents-for-every-matter-tailored-to-you
- Subject: Harvey

Harvey introduced a library of over 500 ready-to-use legal agents built and tested by lawyers, covering practice areas such as M&A, funds, in-house procurement, labor and employment, IP, and white collar investigations. The update also includes an improved Agent Builder that lets teams customize agents in plain English by adjusting scope, steps, and adding organization-specific context and precedents. The upgraded platform is available in early access to select customers.

### ElevenLabs — ElevenLabs crosses $500M ARR as it welcomes new investors including BlackRock, NVIDIA, Jamie Foxx and Eva Longoria

- Company: ElevenLabs (elevenlabs.io)
- Announced: 2026-05-05T12:00:00+00:00
- Category: funding-company-news
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://elevenlabs.io/blog/500m-arr-and-new-investors
- Record: https://forck.live/items/3359-elevenlabs-crosses-500m-arr-as-it-welcomes-new-investors-including-blackrock
- Subject: Voice AI

ElevenLabs reached $500M ARR and added new investors including BlackRock, NVIDIA, Jamie Foxx and Eva Longoria.

### OpenAI — Unlocking large scale AI training networks with MRC (Multipath Reliable Connection)

- Company: OpenAI (openai.com)
- Announced: 2026-05-05T10:00:00+00:00
- Category: infrastructure-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/mrc-supercomputer-networking
- Record: https://forck.live/items/152-unlocking-large-scale-ai-training-networks-with-mrc-multipath-reliable
- Subject: GPT / ChatGPT / API

OpenAI introduces MRC, a new supercomputer networking protocol released via OCP to improve resilience and performance in large-scale AI training clusters.

### OpenAI — GPT-5.5 Instant: smarter, clearer, and more personalized

- Company: OpenAI (openai.com)
- Announced: 2026-05-05T10:00:00+00:00
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://openai.com/index/gpt-5-5-instant
- Record: https://forck.live/items/151-gpt-5-5-instant-smarter-clearer-and-more-personalized
- Subject: GPT / ChatGPT / API

GPT-5.5 Instant is a model update that becomes ChatGPT's default model, offering smarter, more accurate answers, reduced hallucinations, and improved personalization controls.

### Vercel — How KIKO Milano scales for Black Friday

- Company: Vercel (vercel.com)
- Announced: 2026-05-05T07:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/how-kiko-milano-scales-for-black-friday
- Record: https://forck.live/items/20711-how-kiko-milano-scales-for-black-friday
- Subject: v0 / AI SDK / AI Gateway

KIKO Milano's ecommerce team eliminated 3 weeks of manual Black Friday infrastructure prep after migrating to Vercel, reduced app build times by 75%, and now deploys multiple times per day instead of limiting releases. The team estimates saving nearly a full day per week in operational friction.

### Supabase — Realtime or Pipelines? How to choose the right tool

- Company: Supabase (supabase.com)
- Announced: 2026-05-05
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://supabase.com/blog/realtime-or-pipelines-how-to-choose-the-right-tool
- Record: https://forck.live/items/21288-realtime-or-pipelines-how-to-choose-the-right-tool
- Subject: Supabase

Supabase published a guide explaining the difference between Supabase Realtime and Supabase Pipelines, two products that both use logical replication from Postgres but serve different purposes: Realtime delivers live updates to users' browsers and apps, while Pipelines reliably moves data to analytical destinations.

### OpenClaw — OpenClaw Had a Rough Week

- Company: OpenClaw (openclaw.ai)
- Announced: 2026-05-05
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openclaw.ai/blog/openclaw-rough-week/
- Record: https://forck.live/items/17768-openclaw-had-a-rough-week
- Subject: OpenClaw

OpenClaw acknowledged a rough week of instability around April 24–29, 2026, with slower gateways and plugin dependency repair loops. The project is shrinking its core, moving optional components to ClawHub, and plans to announce an LTS release later in May. The post also notes that OpenClaw is transitioning from a founder-driven project to a team-based operation with help from the OpenClaw Foundation and OpenAI.

### Luma — Introducing the Uni-1.1 API: Intelligence You Can Direct. Aesthetic You Can Ship.

- Company: Luma (lumalabs.ai)
- Announced: 2026-05-05
- Category: api-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://lumalabs.ai/news/uni-1-1-api
- Record: https://forck.live/items/7965-introducing-the-uni-1-1-api-intelligence-you-can-direct-aesthetic-you-can-ship
- Subject: Ray / Dream Machine

Luma has opened access to the Uni-1.1 API, a REST interface for its Unified Intelligence model for image generation and natural-language editing. The API offers two endpoints: Generate Image and Modify Image, with support for reference-guided generation and natural-language editing. It is available in two tiers, Build and Scale, at less than half the price and latency of comparable models.

### OpenAI — New ways to buy ChatGPT ads

- Company: OpenAI (openai.com)
- Announced: 2026-05-05
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/new-ways-to-buy-chatgpt-ads
- Record: https://forck.live/items/154-new-ways-to-buy-chatgpt-ads
- Subject: GPT / ChatGPT / API

OpenAI adds a beta self-serve Ads Manager, CPC bidding, and enhanced measurement tools to ChatGPT ads, with privacy protection and separation of conversations from ads.

### OpenAI — Advancing youth safety and wellbeing in EMEA

- Company: OpenAI (openai.com)
- Announced: 2026-05-05
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/advancing-youth-safety-in-emea
- Record: https://forck.live/items/153-advancing-youth-safety-and-wellbeing-in-emea
- Subject: GPT / ChatGPT / API

OpenAI announces European Youth Safety Blueprint and EMEA Youth & Wellbeing Grants to advance safe, responsible AI for teens, families, and educators.

### OpenAI — OpenAI and PwC collaborate to reimagine the office of the CFO

- Company: OpenAI (openai.com)
- Announced: 2026-05-04T21:00:00+00:00
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/openai-pwc-finance-collaboration
- Record: https://forck.live/items/155-openai-and-pwc-collaborate-to-reimagine-the-office-of-the-cfo
- Subject: GPT / ChatGPT / API

OpenAI and PwC announce a partnership to help enterprises use AI agents to automate finance workflows, improve forecasting, strengthen controls, and modernize the CFO function.

### Harvey — Introducing Harvey’s Transformation Office

- Company: Harvey (harvey.ai)
- Announced: 2026-05-04T18:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/introducing-harveys-transformation-office
- Record: https://forck.live/items/17573-introducing-harvey-s-transformation-office
- Subject: Harvey

Harvey introduced the Transformation Office, a service that partners with legal teams to drive AI adoption and deliver measurable business outcomes. The team includes legal industry veterans and change management professionals, and uses structured playbooks covering technology, skills, communications, structures, and governance. The Transformation Office is already working with strategic launch partners and scaling to meet customer demand.

### Vercel — Introducing deepsec: The security harness for finding vulnerabilities in your codebase

- Company: Vercel (vercel.com)
- Announced: 2026-05-04T04:00:00+00:00
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/introducing-deepsec-find-and-fix-vulnerabilities-in-your-code-base
- Record: https://forck.live/items/20713-introducing-deepsec-the-security-harness-for-finding-vulnerabilities-in-your
- Subject: v0 / AI SDK / AI Gateway

Vercel open-sourced deepsec, a security harness that uses coding agents (Claude and Codex) to find vulnerabilities in large codebases. It runs on a user's own infrastructure, supports optional parallel execution via Vercel Sandboxes, and includes a plugin system for custom scanners. The tool has been tested internally and on customer codebases, with a reported false positive rate of roughly 10-20%.

### Vercel — How General Intelligence used agents to build an agent platform on Vercel

- Company: Vercel (vercel.com)
- Announced: 2026-05-04T04:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://vercel.com/blog/how-general-intelligence-used-agents-to-build-an-agent-platform-on-vercel
- Record: https://forck.live/items/20712-how-general-intelligence-used-agents-to-build-an-agent-platform-on-vercel
- Subject: v0 / AI SDK / AI Gateway

General Intelligence, an 8-person company building the Cofounder agent platform, migrated its Python backend from Render to Vercel to achieve full programmatic control of infrastructure via CLI and APIs. The move allows their CTO agent to manage deployments, DNS, billing, and configs end-to-end, enabling engineers to ship an average of 10 PRs and 70+ commits per day with over 4,000 preview branches running at once.

### Supabase — Branching Without Git Is Now The Default

- Company: Supabase (supabase.com)
- Announced: 2026-05-04
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/branching-without-git-is-now-the-default
- Record: https://forck.live/items/21289-branching-without-git-is-now-the-default
- Subject: Supabase

Supabase made dashboard-based branching the default for all projects, removing the requirement for a GitHub connection. Users can create branches, make schema changes, review diffs, and merge directly from the Supabase Dashboard. The new pg-delta diffing engine powers the merge experience, and the git-based workflow remains fully supported.

### Kyutai — Pocket TTS now supports six languages

- Company: Kyutai (kyutai.org)
- Announced: 2026-05-04
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://kyutai.org/blog/2026-05-04-pocket-tts-multilingual
- Record: https://forck.live/items/18343-pocket-tts-now-supports-six-languages
- Subject: Moshi / Pocket TTS

Pocket TTS now speaks English, French, German, Spanish, Portuguese, and Italian.

### Runway — Building Real-Time Video Agent from a Single Image with Runway Characters

- Company: Runway (runway.com)
- Announced: 2026-05-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://runway.com/news/engineering/building-runway-characters
- Record: https://forck.live/items/7887-building-real-time-video-agent-from-a-single-image-with-runway-characters
- Subject: Gen / Aleph

Runway announced Runway Characters, a real-time video agent that transforms a single reference image into a conversational video character at 24fps with low latency. It is built on the GWM-1 model and requires no fine-tuning. The system achieves 37ms model time per frame and 1.75 seconds server-side turn-around.

### Runway — 60x Faster Cold Starts: Treating Peer GPUs as Weight Servers

- Company: Runway (runway.com)
- Announced: 2026-05-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://runway.com/news/engineering/60x-faster-cold-starts-treating-peer-gpus-as-weight-servers
- Record: https://forck.live/items/7886-60x-faster-cold-starts-treating-peer-gpus-as-weight-servers
- Subject: Gen / Aleph

Runway built a system called NCCLBack that reduces model cold-start times from minutes to seconds by broadcasting weights between GPUs instead of downloading them from storage.

### Amp — GPT-5.5 In Deep

- Company: Amp (ampcode.com)
- Announced: 2026-05-04
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://ampcode.com/news/gpt-5.5
- Record: https://forck.live/items/7750-gpt-5-5-in-deep
- Subject: Amp

Amp has updated its deep mode to run on GPT-5.5, which is a better coding agent than GPT-5.4 with improved steerability, interactivity, and constraint adherence. The default reasoning effort in deep mode was changed from high to medium because higher effort performed worse in evaluations.

### Cursor — Model controls, spend management, and usage analytics

- Company: Cursor (cursor.com)
- Announced: 2026-05-04
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/05-04-26
- Record: https://forck.live/items/2614-model-controls-spend-management-and-usage-analytics
- Subject: Cursor

Cursor introduces new model access controls, soft spend limits with automatic alerts, and updated usage analytics for Enterprise admins.

### Together AI — Foundational research powering efficient inference at scale

- Company: Together AI (together.ai)
- Announced: 2026-05-04
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.together.ai/blog/foundational-research-powering-efficient-inference-at-scale
- Record: https://forck.live/items/2351-foundational-research-powering-efficient-inference-at-scale
- Subject: Inference platform

Together AI discusses its approach to efficient inference at scale, highlighting research contributions like FlashAttention-4, ThunderKittens, and Aurora, and its full-stack hardware optimization on NVIDIA Blackwell hardware. The post covers the economics of inference, but does not announce a new product, model, or specific pricing change; it is a general thought-leadership piece about inference infrastructure.

### OpenAI — How OpenAI delivers low-latency voice AI at scale

- Company: OpenAI (openai.com)
- Announced: 2026-05-04
- Category: infrastructure-release
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/delivering-low-latency-voice-ai-at-scale
- Record: https://forck.live/items/156-how-openai-delivers-low-latency-voice-ai-at-scale
- Subject: GPT / ChatGPT / API

OpenAI rebuilt its WebRTC stack to enable low-latency, globally scalable real-time voice AI with seamless conversational turn-taking.