AI News for July 3, 2026 — 7min.ai

[NEWS] Altman's proposed OpenAI government stake: $42.6B via Alaska-style fund: Sam Altman personally proposed the terms for the US government's roughly 5% equity stake in OpenAI, worth about $42.6 billion at the company's $852 billion valuation, suggesting the holding be routed through a sovereign-wealth-style vehicle modeled on the Alaska Permanent Fund, according to the Financial Times.
[NEWS] Anthropic in talks with Samsung to build a custom AI chip: Anthropic is discussing a custom AI chip with Samsung Electronics as a manufacturing partner, according to The Information.
[NEWS] Microsoft launches $2.5B Frontier Company to embed 6,000 AI engineers in enterprises: Microsoft created a new operating unit called Frontier Company, committing $2.5 billion and 6,000 industry and engineering experts to deploy AI directly inside customer organizations.
[TOOLS] Z.ai's free ZCode expands to Windows and Linux, now also targets Copilot and Antigravity: Z.ai's ZCode agentic coding tool, built for GLM-5.2, is free to use and now available on Windows and Linux in addition to macOS, with bring-your-own-key support for other models.
[RESEARCH] Alibaba's SkillWeaver framework cuts AI agent token use by 99%: Alibaba researchers introduced SkillWeaver, a framework that builds an execution graph for a task and routes each step to the right tool rather than exposing an agent to its entire tool library.
[RESEARCH] Fable 5 hits record 16% on freelance jobs benchmark, up from 2.5% eight months ago: The Remote Labor Index, which measures how often AI agents complete real paid freelance projects at professional quality, now shows Anthropic's Fable 5 automating 16.1% of projects, the highest score ever recorded.
[NEWS] ElevenLabs in talks for tender offer at $22B valuation: ElevenLabs has held early talks with investors on a secondary offering that would let employees sell shares and value the AI voice startup at roughly $22 billion.
[RESEARCH] Study finds a single transformer layer can match full-parameter RL fine-tuning: Researchers found that training just one transformer layer during RL post-training can match the performance gains of updating all parameters.
[OPINION] Sam Altman calls for a US-led global AI forum as ChatGPT loses ground: OpenAI CEO Sam Altman used a Financial Times op-ed to propose a US-led international forum that would set standards for AI and make the technology available to nations and companies that follow the rules.
[NEWS] Zuckerberg tells staff Meta's AI agents are progressing slower than hoped: Mark Zuckerberg told employees at an internal town hall that AI agent development has not accelerated the way Meta expected, and admitted this year's cuts of roughly 8,000 jobs plus 7,000 reassignments were not as 'clean' as they should have been.

About 7min.ai · Subscribe to newsletter