# forck.live — 6–12 April 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-W15, 6–12 April 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-W15
- The current issue, and the catalogue of the other weeks: https://forck.live/briefing.md
- Catalogue: https://forck.live/llms.txt

## Announcements

### fal — Introducing PATINA

- Company: fal (fal.ai)
- Announced: 2026-04-10T16:44:36+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://blog.fal.ai/introducing-patina/
- Record: https://forck.live/items/7999-introducing-patina
- Subject: fal platform

fal introduced PATINA, a model that generates high-resolution PBR maps from rendered images, bridging the gap between AI-generated images and traditional CGI workflows. PATINA uses a modified FLUX.2 klein backbone with a DINOv2 adapter and was trained on CC0 material libraries with a custom Cook-Torrance BRDF renderer. The model outputs basecolor, normal, roughness, metalness, and height maps, and is available via endpoints starting at $0.08 for a full material set.

### Factory AI — How Missions Work

- Company: Factory AI (factory.com)
- Announced: 2026-04-10T07:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://factory.com/news/missions-architecture
- Record: https://forck.live/items/17860-how-missions-work
- Subject: Droid

Factory AI's Missions system breaks large software projects into focused units handled by fresh agents with narrowly scoped goals, shared state, and explicit validation. It separates roles into orchestrator, workers, and validators, and uses test-driven development at two levels. The system also allows model specialization per role.

### OpenAI — Prompting fundamentals

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/prompting
- Record: https://forck.live/items/11722-prompting-fundamentals
- Subject: GPT / ChatGPT / API

OpenAI published a guide covering prompting fundamentals and how to write clear, effective prompts to improve responses from ChatGPT.

### Runway — Runway Expands Collaboration with NYU Tisch School of the Arts

- Company: Runway (runway.com)
- Announced: 2026-04-10
- Category: partnership-acquisition
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://runway.com/news/runway-expands-collaboration-with-nyu-tisch-school-of-the-arts
- Record: https://forck.live/items/7889-runway-expands-collaboration-with-nyu-tisch-school-of-the-arts
- Subject: Gen / Aleph

Runway deepened its collaboration with NYU Tisch School of the Arts, providing access to its full suite of AI tools to students in the Interactive Telecommunications Program (ITP), Interactive Media Arts (IMA), and the Hyper Cinema Lab.

### Anthropic — Preparing your security program for AI-accelerated offense

- Company: Anthropic (anthropic.com)
- Announced: 2026-04-10
- Category: safety-policy-update
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://claude.com/blog/preparing-your-security-program-for-ai-accelerated-offense
- Record: https://forck.live/items/7410-preparing-your-security-program-for-ai-accelerated-offense
- Subject: Claude

Anthropic published security recommendations based on its findings to help organizations prepare for AI-accelerated offense, including prioritizing patching and preparing for increased vulnerability reports.

### OpenAI — Learn ChatGPT workflows for sales teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/sales
- Record: https://forck.live/items/4151-learn-chatgpt-workflows-for-sales-teams
- Subject: GPT / ChatGPT / API

OpenAI offers a course on how sales teams can use ChatGPT for account research, personalized outreach, deal management, and pipeline performance improvement.

### OpenAI — Learn ChatGPT workflows for marketing teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/marketing
- Record: https://forck.live/items/4150-learn-chatgpt-workflows-for-marketing-teams
- Subject: GPT / ChatGPT / API

OpenAI Academy offers a learning resource on how marketing teams can use ChatGPT for campaign planning, content generation, performance analysis, and faster execution.

### OpenAI — Learn ChatGPT workflows for finance teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/finance
- Record: https://forck.live/items/4149-learn-chatgpt-workflows-for-finance-teams
- Subject: GPT / ChatGPT / API

OpenAI offers a learning resource on how finance teams can use ChatGPT for reporting, data analysis, forecasting, and communication.

### OpenAI — Our response to the Axios developer tool compromise

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/axios-developer-tool-compromise
- Record: https://forck.live/items/208-our-response-to-the-axios-developer-tool-compromise
- Subject: GPT / ChatGPT / API

OpenAI responded to the Axios supply chain attack by rotating macOS code signing certificates, updating apps, and confirming that no user data was compromised.

### OpenAI — Creating images with ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/image-generation
- Record: https://forck.live/items/207-creating-images-with-chatgpt
- Subject: GPT / ChatGPT / API

OpenAI published a tutorial on creating and refining images using ChatGPT, focusing on prompt clarity and iteration.

### OpenAI — Healthcare

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/healthcare
- Record: https://forck.live/items/206-healthcare
- Subject: GPT / ChatGPT / API

This is about how clinicians can use ChatGPT for healthcare tasks such as diagnosis, documentation, and patient care, with HIPAA-compliant tools.

### OpenAI — Learn ChatGPT for Sales Teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/sales
- Record: https://forck.live/items/205-learn-chatgpt-for-sales-teams
- Subject: GPT / ChatGPT / API

OpenAI offers a learning resource on how sales teams can use ChatGPT for account research, personalized outreach, deal management, and pipeline performance improvement.

### OpenAI — ChatGPT for customer success teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/customer-success
- Record: https://forck.live/items/204-chatgpt-for-customer-success-teams
- Subject: GPT / ChatGPT / API

This is an educational page about how customer success teams can use ChatGPT to manage accounts, improve communication, reduce churn, and increase adoption and renewals.

### OpenAI — Analyzing data with ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/data-analysis
- Record: https://forck.live/items/203-analyzing-data-with-chatgpt
- Subject: GPT / ChatGPT / API

This post teaches how to use ChatGPT's data analysis features to explore datasets, generate insights, create visualizations, and turn findings into decisions.

### OpenAI — ChatGPT for operations teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/operations
- Record: https://forck.live/items/202-chatgpt-for-operations-teams
- Subject: GPT / ChatGPT / API

OpenAI published an educational resource titled 'ChatGPT for operations teams' that teaches how operations teams can use ChatGPT to streamline workflows, improve coordination, standardize processes, and drive faster execution.

### OpenAI — AI fundamentals

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/what-is-ai
- Record: https://forck.live/items/201-ai-fundamentals
- Subject: GPT / ChatGPT / API

This is an introductory educational guide that explains what AI is, how it works, and how tools like ChatGPT use large language models.

### OpenAI — ChatGPT for marketing teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/marketing
- Record: https://forck.live/items/200-chatgpt-for-marketing-teams
- Subject: GPT / ChatGPT / API

OpenAI provides a learning resource on how marketing teams can use ChatGPT for campaign planning, content generation, performance analysis, and faster execution.

### OpenAI — Using custom GPTs

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/custom-gpts
- Record: https://forck.live/items/199-using-custom-gpts
- Subject: GPT / ChatGPT / API

This is a learning resource about building and using custom GPTs to automate workflows, maintain consistent outputs, and create purpose-built AI assistants.

### OpenAI — ChatGPT for finance teams

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/finance
- Record: https://forck.live/items/198-chatgpt-for-finance-teams
- Subject: GPT / ChatGPT / API

OpenAI offers a learning resource on how finance teams can use ChatGPT for reporting, data analysis, forecasting, and communication.

### OpenAI — Financial services

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/financial-services
- Record: https://forck.live/items/197-financial-services
- Subject: GPT / ChatGPT / API

OpenAI provides AI resources for financial services, including prompt packs, GPTs, guides, and tools to help institutions deploy and scale AI securely.

### OpenAI — Working with files in ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/academy/working-with-files
- Record: https://forck.live/items/196-working-with-files-in-chatgpt
- Subject: GPT / ChatGPT / API

OpenAI explains how to use file uploading and interaction features within ChatGPT for tasks such as data analysis, document summarization, and content generation from PDFs and spreadsheets.

### OpenAI — Responsible and safe use of AI

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/responsible-and-safe-use
- Record: https://forck.live/items/195-responsible-and-safe-use-of-ai
- Subject: GPT / ChatGPT / API

This is an educational resource on responsible AI use and best practices for safety, accuracy, and transparency when using tools like ChatGPT.

### OpenAI — Brainstorming with ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/brainstorming
- Record: https://forck.live/items/194-brainstorming-with-chatgpt
- Subject: GPT / ChatGPT / API

A guide on using ChatGPT for brainstorming, organizing thinking, and structuring rough concepts into actionable plans.

### OpenAI — Research with ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/search-and-deep-research
- Record: https://forck.live/items/193-research-with-chatgpt
- Subject: GPT / ChatGPT / API

A guide on using ChatGPT's search and deep research capabilities for research purposes.

### OpenAI — Applications of AI at OpenAI

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/applications-of-ai
- Record: https://forck.live/items/192-applications-of-ai-at-openai
- Subject: GPT / ChatGPT / API

OpenAI's academy page describes how its products ChatGPT, Codex, and APIs are used in real-world applications for work, development, and everyday tasks.

### OpenAI — ChatGPT for managers

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/managers
- Record: https://forck.live/items/191-chatgpt-for-managers
- Subject: GPT / ChatGPT / API

OpenAI offers a learning resource for managers on using ChatGPT to prepare for conversations, write clear feedback, stay organized, and improve team effectiveness.

### OpenAI — Using projects in ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/projects
- Record: https://forck.live/items/190-using-projects-in-chatgpt
- Subject: GPT / ChatGPT / API

This post teaches users how to organize chats, files, and instructions using the projects feature in ChatGPT, and how to manage ongoing work and collaborate more effectively.

### OpenAI — Using skills

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: capability-change
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/skills
- Record: https://forck.live/items/189-using-skills
- Subject: GPT / ChatGPT / API

This post teaches users how to create and use ChatGPT skills for building reusable workflows, automating recurring tasks, and ensuring consistent outputs.

### OpenAI — Personalizing ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/personalization
- Record: https://forck.live/items/188-personalizing-chatgpt
- Subject: GPT / ChatGPT / API

The page teaches users how to personalize ChatGPT with custom instructions and memory for more tailored responses.

### OpenAI — Writing with ChatGPT

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/writing
- Record: https://forck.live/items/187-writing-with-chatgpt
- Subject: GPT / ChatGPT / API

This is an educational resource teaching how to use ChatGPT for writing tasks.

### OpenAI — ChatGPT for research

- Company: OpenAI (openai.com)
- Announced: 2026-04-10
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/academy/research
- Record: https://forck.live/items/186-chatgpt-for-research
- Subject: GPT / ChatGPT / API

This is a guide on how to use ChatGPT for research, including gathering sources, analyzing information, and creating structured insights with citations.

### Harvey — The Brief: April 2026

- Company: Harvey (harvey.ai)
- Announced: 2026-04-09T15:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.harvey.ai/blog/the-brief-april-2026
- Record: https://forck.live/items/17585-the-brief-april-2026
- Subject: Harvey

Harvey announced a range of new capabilities for its AI-powered legal platform, including the availability of OpenAI's GPT-5.4 model across Assistant, Vault, and Agent Builder. Users can now edit Word files directly in Assistant, batch-edit multiple documents, and sync folders from document management systems to Vault. The update also introduces a Box integration, custom writing styles, improved Magic Prompt, and expanded legal research sources.

### ElevenLabs — Enterprise voice AI, deployed locally

- Company: ElevenLabs (elevenlabs.io)
- Announced: 2026-04-09T12:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://elevenlabs.io/blog/enterprise-voice-ai-deployed-locally
- Record: https://forck.live/items/3370-enterprise-voice-ai-deployed-locally
- Subject: Voice AI

ElevenLabs announces enterprise voice AI that can be deployed locally.

### Google Research — ConvApparel: Measuring and bridging the realism gap in user simulators

- Company: Google Research (research.google)
- Announced: 2026-04-09T11:22:00+00:00
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/convapparel-measuring-and-bridging-the-realism-gap-in-user-simulators/
- Record: https://forck.live/items/1348-convapparel-measuring-and-bridging-the-realism-gap-in-user-simulators
- Subject: Research

Google Research introduces ConvApparel, a dataset and evaluation framework to measure and bridge the realism gap in LLM-based user simulators, using a dual-agent protocol and counterfactual validation.

### Vercel — Agentic Infrastructure

- Company: Vercel (vercel.com)
- Announced: 2026-04-09T07:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/agentic-infrastructure
- Record: https://forck.live/items/20718-agentic-infrastructure
- Subject: v0 / AI SDK / AI Gateway

Vercel describes its platform as 'Agentic Infrastructure' designed for software built and operated by coding agents. The post details three areas: infrastructure for coding agents to deploy to, infrastructure for building and running agents (including AI SDK, Chat SDK, AI Gateway, Fluid compute, Workflows, Queues, Sandbox, and Observability), and infrastructure that itself is agentic, capable of autonomously investigating anomalies and proposing fixes. Vercel reports that over 50% of deployments on its platform are now initiated by coding agents as of September 2026.

### Supabase — AI Agents Know About Supabase. They Don't Always Use It Right.

- Company: Supabase (supabase.com)
- Announced: 2026-04-09
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/supabase-agent-skills
- Record: https://forck.live/items/21293-ai-agents-know-about-supabase-they-don-t-always-use-it-right
- Subject: Supabase

Supabase released Supabase Agent Skills, an open-source set of instructions that teach AI coding agents how to build on Supabase correctly. The skills cover documentation access, security, tooling workflow, and schema management, and can be installed with a single command for agents like Claude Code, Codex, GitHub Copilot, and Cursor.

### Gradium — Evaluating Phonon: how we made the best TTS model for edge devices

- Company: Gradium (gradium.ai)
- Announced: 2026-04-09
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://gradium.ai/blog/evaluating-phonon
- Record: https://forck.live/items/18323-evaluating-phonon-how-we-made-the-best-tts-model-for-edge-devices
- Subject: Gradium TTS / Phonon

Gradium's on-device TTS model Phonon (≈100M parameters) achieved a 1.48% word error rate and 56.37% speaker similarity on the Seed-TTS English benchmark, outperforming models 2–5× its size. An update notes that as of May 2026, Phonon reaches 1.00% WER on the same benchmark. Phonon supports voice cloning from a 10-second reference sample and is currently in private beta.

### Luma — Luma Submits 21 AI-Generated Finalists to Cannes Lions After Inspiring 400 Ads in Less Than 8 Weeks

- Company: Luma (lumalabs.ai)
- Announced: 2026-04-09
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://lumalabs.ai/news/luma-dreambrief-submission-finalists
- Record: https://forck.live/items/7968-luma-submits-21-ai-generated-finalists-to-cannes-lions-after-inspiring-400-ads
- Subject: Ray / Dream Machine

Luma submitted 21 AI-generated finalist submissions from its Dream Brief competition to the Cannes Lions International Festival of Creativity. The competition generated nearly 400 submissions in under two months, using Luma's generative AI platform. Luma will sponsor the entries and award $1,000,000 to any creator whose submission wins a Gold Lion.

### Cognition — Launching in Japan with Takumi Masai

- Company: Cognition (cognition.com)
- Announced: 2026-04-09
- Category: availability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cognition.com/blog/launching-in-japan
- Record: https://forck.live/items/7671-launching-in-japan-with-takumi-masai
- Subject: Devin / SWE

Cognition is expanding into Japan, marking its first step into Asia, and has appointed Takumi Masai as Japan President and General Manager. The company is building a local team to work with Japanese enterprises and partners, and is already working with organizations such as DeNA and Mizuho Securities.

### LG AI Research — LG's first publicly released Vision Language Model, EXAONE 4.5

- Company: LG AI Research (lgresearch.ai)
- Announced: 2026-04-09
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.lgresearch.ai/blog/view?seq=641
- Record: https://forck.live/items/4752-lg-s-first-publicly-released-vision-language-model-exaone-4-5
- Subject: EXAONE

LG AI Research announces EXAONE 4.5, its first vision language model released under an open-weight approach, featuring a native multimodal pretraining approach and improved inference speed via multi-token prediction.

### Hugging Face — Multimodal Embedding & Reranker Models with Sentence Transformers

- Company: Hugging Face (huggingface.co)
- Announced: 2026-04-09
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://huggingface.co/blog/multimodal-sentence-transformers
- Record: https://forck.live/items/1522-multimodal-embedding-reranker-models-with-sentence-transformers
- Subject: Platform

Sentence Transformers v5.4 adds multimodal embedding and reranker support, enabling encoding and comparison of text, images, audio, and video with the same API. The update allows cross-modal similarity and retrieval, demonstrated with Qwen3-VL-Embedding-2B.

### Hugging Face — Waypoint-1.5: Higher-Fidelity Interactive Worlds for Everyday GPUs

- Company: Hugging Face (huggingface.co)
- Announced: 2026-04-09
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://huggingface.co/blog/waypoint-1-5
- Record: https://forck.live/items/1521-waypoint-1-5-higher-fidelity-interactive-worlds-for-everyday-gpus
- Subject: Platform

Waypoint-1.5 is Overworld's next real-time video world model with improved visual fidelity, a 360p tier for broader hardware support, and trained on nearly 100x more data than Waypoint-1.

### OpenAI — CyberAgent moves faster with ChatGPT Enterprise and Codex

- Company: OpenAI (openai.com)
- Announced: 2026-04-09
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/cyberagent
- Record: https://forck.live/items/209-cyberagent-moves-faster-with-chatgpt-enterprise-and-codex
- Subject: GPT / ChatGPT / API

CyberAgent is using ChatGPT Enterprise and Codex to securely scale AI adoption, improve quality, and accelerate decisions across advertising, media, and gaming.

### Google Research — Improving the academic workflow: Introducing two AI agents for better figures and peer review

- Company: Google Research (research.google)
- Announced: 2026-04-08T20:01:33+00:00
- Category: developer-tool-release
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://research.google/blog/improving-the-academic-workflow-introducing-two-ai-agents-for-better-figures-and-peer-review/
- Record: https://forck.live/items/1349-improving-the-academic-workflow-introducing-two-ai-agents-for-better-figures
- Subject: Research

Google Research introduces two AI agents, PaperVizAgent for generating academic figures and ScholarPeer for automated peer review, both outperforming existing baselines.

### ByteDance — Introducing Seed Full-Duplex Speech LLM: Attentive Listening, Robust Interference Suppression, Enabling More Natural Interaction

- Company: ByteDance (bytedance.com)
- Announced: 2026-04-08T16:00:00+00:00
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://seed.bytedance.com/en/blog/introducing-seed-full-duplex-speech-llm-attentive-listening-robust-interference-suppression-enabling-more-natural-interaction
- Record: https://forck.live/items/4665-introducing-seed-full-duplex-speech-llm-attentive-listening-robust
- Subject: Seed / Doubao

ByteDance introduces Seeduplex, a native full-duplex speech LLM that enables simultaneous listening and speaking, achieving breakthroughs in interference suppression and adaptive endpoint detection. The model has been fully rolled out on the Doubao App, providing a more natural and fluent voice interaction experience.

### Harvey — Harvey in Practice: How Emerging Companies and VC Teams Support Growth at Every Stage

- Company: Harvey (harvey.ai)
- Announced: 2026-04-08T15:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/harvey-in-practice-emerging-companies-and-vc-teams
- Record: https://forck.live/items/17586-harvey-in-practice-how-emerging-companies-and-vc-teams-support-growth-at-every
- Subject: Harvey

Harvey's guide shows how emerging company and venture capital lawyers can use Harvey across the company lifecycle, from client intake and formation through financings and corporate work, to standardize workflows, leverage firm knowledge, and accelerate review, drafting, and communication with founders.

### OpenAI — The next phase of enterprise AI

- Company: OpenAI (openai.com)
- Announced: 2026-04-08T14:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/next-phase-of-enterprise-ai
- Record: https://forck.live/items/210-the-next-phase-of-enterprise-ai
- Subject: GPT / ChatGPT / API

OpenAI outlines the next phase of enterprise AI, as adoption accelerates across industries with Frontier, ChatGPT Enterprise, Codex, and company-wide AI agents.

### Stability AI — Introducing Brand Studio: The creative production platform powered by your brand

- Company: Stability AI (stability.ai)
- Announced: 2026-04-08T12:57:16+00:00
- Category: product-launch
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://stability.ai/news-updates/brand-studio-by-stability-ai-creative-production-platform-for-brands
- Record: https://forck.live/items/17476-introducing-brand-studio-the-creative-production-platform-powered-by-your-brand
- Subject: Stable Diffusion

Stability AI launched Brand Studio, an end-to-end creative production platform for professional teams. The platform includes Brand Central for custom Brand ID models and Campaigns, Producer Mode for step-by-step production plans, Curated Model Routing that selects models based on marketing requirements, and precision editing tools such as Precision Inpainting and Product Insertion. It also offers enterprise features like SSO and access controls.

### Vercel — Zero Data Retention on AI Gateway

- Company: Vercel (vercel.com)
- Announced: 2026-04-08T07:00:00+00:00
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://vercel.com/blog/zdr-on-ai-gateway
- Record: https://forck.live/items/20719-zero-data-retention-on-ai-gateway
- Subject: v0 / AI SDK / AI Gateway

Vercel expanded AI Gateway's compliance capabilities with team-wide Zero Data Retention (ZDR), allowing teams to enforce strict data policies from the dashboard without code changes. The update also includes per-request ZDR for specific sensitive workflows and explicit controls to disallow prompt training. Team-wide ZDR is available for Pro and Enterprise teams.

### OpenAI — Introducing the Child Safety Blueprint

- Company: OpenAI (openai.com)
- Announced: 2026-04-08T05:00:00+00:00
- Category: safety-policy-update
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://openai.com/index/introducing-child-safety-blueprint
- Record: https://forck.live/items/211-introducing-the-child-safety-blueprint
- Subject: GPT / ChatGPT / API

OpenAI introduces a Child Safety Blueprint, a roadmap for responsible AI development with safeguards, age-appropriate design, and collaboration to protect young people online.

### Supabase — Custom OIDC Providers for Supabase Auth

- Company: Supabase (supabase.com)
- Announced: 2026-04-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://supabase.com/blog/custom-oauth-oidc-providers
- Record: https://forck.live/items/21294-custom-oidc-providers-for-supabase-auth
- Subject: Supabase

Supabase launched Custom OIDC Providers for its Auth service, enabling users to connect any standards-compliant OpenID Connect identity provider. The feature supports PKCE by default, custom authorization parameters, multi-platform client IDs, and email-optional providers. It is manageable via the Dashboard and Admin API, with a limit of three custom providers per project.

### Genspark — Introducing Genspark AI Workspace 4.0: Your AI Employee, Now Everywhere

- Company: Genspark (genspark.ai)
- Announced: 2026-04-08
- Category: not stated
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://www.genspark.ai/blog/genspark-ai-workspace-4
- Record: https://forck.live/items/18257-introducing-genspark-ai-workspace-4-0-your-ai-employee-now-everywhere
- Subject: Genspark

### Factory AI — The Factory Desktop App

- Company: Factory AI (factory.com)
- Announced: 2026-04-08
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://factory.com/news/factory-desktop
- Record: https://forck.live/items/17840-the-factory-desktop-app
- Subject: Droid

Factory released a desktop app for macOS and Windows that provides a native interface for running multiple Droids simultaneously. The app introduces Droid Computers with persistent storage, SSH access, and instant resume, and supports BYO Machine for running on local hardware. It also enables computer use, allowing Droids to control other applications like VS Code and browsers, and includes VS Code integration, AI-native visualization, and mobile access.

### Liquid AI — LFM2.5-VL-450M: Structured Visual Intelligence, Edge to Cloud

- Company: Liquid AI (liquid.ai)
- Announced: 2026-04-08
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.liquid.ai/blog/lfm2-5-vl-450m
- Record: https://forck.live/items/16784-lfm2-5-vl-450m-structured-visual-intelligence-edge-to-cloud
- Subject: LFM / d1 models

Liquid AI released LFM2.5-VL-450M, an improved version of LFM2-VL-450M with added grounding capabilities, better instruction following, and function calling support. The model was pre-trained on 28T tokens (up from 10T) and uses preference optimization and reinforcement learning to improve multimodal behavior. It achieves 81.28 on RefCOCO-M for bounding box prediction and runs on edge hardware such as Jetson Orin, Samsung S25 Ultra, and AMD Ryzen AI Max+ 395.

### Cognition — How Devin Is Modernizing COBOL at Fortune 500 Companies

- Company: Cognition (cognition.com)
- Announced: 2026-04-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cognition.com/blog/how-devin-is-modernizing-cobol-at-fortune-500-companies
- Record: https://forck.live/items/7672-how-devin-is-modernizing-cobol-at-fortune-500-companies
- Subject: Devin / SWE

Cognition's AI coding agent Devin has been deployed at several Fortune 500 companies to modernize COBOL systems, including documentation, migration to AWS Lambda, and refactoring. The post describes challenges unique to COBOL and outlines where agents can currently succeed (batch workloads and documentation) versus where full automation remains out of reach (transactional workloads).

### Cursor — Bugbot Learned Rules and MCP Support

- Company: Cursor (cursor.com)
- Announced: 2026-04-08
- Category: capability-change
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cursor.com/changelog/04-08-26
- Record: https://forck.live/items/2622-bugbot-learned-rules-and-mcp-support
- Subject: Cursor

Bugbot now learns from PR feedback to create and auto-promote rules, supports MCP servers on Teams and Enterprise plans, and achieves a 78% resolution rate with a new Fix All action.

### Hugging Face — Safetensors is Joining the PyTorch Foundation

- Company: Hugging Face (huggingface.co)
- Announced: 2026-04-08
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://huggingface.co/blog/safetensors-joins-pytorch-foundation
- Record: https://forck.live/items/1523-safetensors-is-joining-the-pytorch-foundation
- Subject: Platform

Safetensors, a format for storing model weights, is joining the PyTorch Foundation as a hosted project under the Linux Foundation, with governance moving to the community while Hugging Face maintainers continue leading. No breaking changes for users.

### Harvey — Building Spectre

- Company: Harvey (harvey.ai)
- Announced: 2026-04-07T17:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.harvey.ai/blog/building-spectre-internal-collaborative-cloud-agent-platform
- Record: https://forck.live/items/17587-building-spectre
- Subject: Harvey

Harvey describes Spectre, its internal collaborative cloud agent platform that turns requests from Slack, the web app, or automations into durable runs executed in isolated sandboxes, connected to systems like GitHub, Datadog, and Linear, and returning reviewable artifacts such as summaries, diffs, branches, and pull requests. The platform uses durable run records and ephemeral workers to enable sharing, resumability, and collaboration across engineers, product managers, and designers.

### Cognition — Introducing SWE 1.6: Improving Model UX

- Company: Cognition (cognition.com)
- Announced: 2026-04-07
- Category: model-update
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://cognition.com/blog/swe-1-6
- Record: https://forck.live/items/7673-introducing-swe-1-6-improving-model-ux
- Subject: Devin / SWE

Cognition released SWE-1.6, a new model for software engineering agents, optimized for intelligence and model UX. It is available in Windsurf with a free tier for the next 3 months and offers speeds up to 950 tok/s for paying users. The model improves on behavioral issues like overthinking and looping compared to the preview version.

### Z.ai — GLM-5.1

- Company: Z.ai (z.ai)
- Announced: 2026-04-07
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://docs.z.ai/release-notes/new-released#2026-04-07
- Record: https://forck.live/items/4448-glm-5-1
- Subject: GLM

Z.ai announced GLM-5.1, a model designed for long-horizon tasks that can work independently for up to 8 hours, with stronger engineering intelligence and capability alignment with Claude Opus 4.6.

### Together AI — What is an AI Native Cloud?

- Company: Together AI (together.ai)
- Announced: 2026-04-07
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.together.ai/blog/what-is-an-ai-native-cloud
- Record: https://forck.live/items/2360-what-is-an-ai-native-cloud
- Subject: Inference platform

Together AI defines an AI Native Cloud as a purpose-built cloud for AI-native companies, describing its key characteristics such as full AI stack integration, fast path from research to production, and support for rapid iteration and scaling.

### OpenAI — Announcing the OpenAI Safety Fellowship

- Company: OpenAI (openai.com)
- Announced: 2026-04-06T10:00:00+00:00
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://openai.com/index/introducing-openai-safety-fellowship
- Record: https://forck.live/items/212-announcing-the-openai-safety-fellowship
- Subject: GPT / ChatGPT / API

OpenAI announces a pilot program for a Safety Fellowship to support independent safety and alignment research and develop next-generation talent.

### LG AI Research — LG Aimers 8th: Taking on the Challenge with ‘EXAONE,’ Korea’s Representative AI

- Company: LG AI Research (lgresearch.ai)
- Announced: 2026-04-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.lgresearch.ai/news/view?seq=639
- Record: https://forck.live/items/10992-lg-aimers-8th-taking-on-the-challenge-with-exaone-korea-s-representative-ai
- Subject: EXAONE

LG AI Research held the 8th LG Aimers offline hackathon, where 27 teams of young participants competed to develop compressed versions of the EXAONE model. The winning team, Hand Cream Mice, took the Grand Prize by using W8A8 quantization to maximize memory efficiency. The event also included a job fair with eight LG affiliates and career support activities.

### LG AI Research — LG Aimers 8th: Taking on the Challenge with ‘EXAONE,’ Korea’s Representative AI

- Company: LG AI Research (lgresearch.ai)
- Announced: 2026-04-06
- Category: not stated
- Coverage: not counted
- Announcement: no
- Group: routine
- Source: https://www.lgresearch.ai/blog/view?seq=639
- Record: https://forck.live/items/4753-lg-aimers-8th-taking-on-the-challenge-with-exaone-korea-s-representative-ai
- Subject: EXAONE

LG AI Research hosted the 8th LG Aimers Hackathon, where 27 teams of young talents competed to develop compressed versions of the EXAONE model, focusing on on-device AI optimization. The event included a job fair with LG affiliates and awarded prizes for the best compressed models.