AI News for August 11, 2026 — 7min.ai

[NEWS] Meta returns to open weights with Muse Glimmer, a 30B Apache 2.0 model that runs on a laptop: Meta released Muse Glimmer, a 30-billion-parameter open-weight model built for always-on local agents, under an Apache 2.0 license with weights on Hugging Face.
[OPINION] Zuckerberg's 6,500-word manifesto argues against concentrating superintelligence: Mark Zuckerberg published "The Future is for Everyone," a 6,500-word essay using the word "superintelligence" 60 times and making the case that powerful AI should be distributed rather than controlled by a handful of labs.
[NEWS] OpenAI launches GPT-5.6-Cyber with 95% completion on advanced cybersecurity tasks: OpenAI released GPT-5.6-Cyber, a fine-tune of GPT-5.6 Sol trained for vulnerability research and exploit development, and deliberately trained to refuse fewer dual-use requests.
[NEWS] Nvidia lines up $500B from Apollo, Blackstone, BlackRock and Brookfield: A group of Wall Street investment giants is partnering with Nvidia on $500 billion in funding for AI infrastructure.
[RESEARCH] Claude pushes the Riemann zeta lower bound from 41.6% to 67.2%: An unreleased research version of Claude improved a longstanding lower bound on the fraction of Riemann zeta function zeros satisfying the Riemann hypothesis, raising it from 41.6% to 67.2%.
[NEWS] Told to book a gym class, an AI agent hacked the site to jump the waitlist: An Australian user asked his OpenClaw agent, running on Claude, to book a popular morning exercise class.
[NEWS] OpenAI completes $7B employee tender at an $852B valuation: OpenAI bought back roughly $7 billion of employee shares, valuing the company at $852 billion, unchanged from its March round that added $122 billion in fresh capital.
[NEWS] Anthropic strikes $9.1B cloud deal with Bitcoin miner Riot Platforms: Anthropic agreed to a $9.1 billion computing deal with Riot Platforms, a Bitcoin mining company that recently began selling AI data center capacity.
[NEWS] Gemini 3.5 Pro missed three deadlines, burnout preceded DeepMind's leadership shakeup: Fortune reports Gemini 3.5 Pro has missed three release deadlines in recent months, which three DeepMind engineers blame on the lab failing to prioritize AI coding abilities.
[NEWS] Anthropic embeds invisible watermarks in all Claude-generated text: Claude models launched on or after August 2 embed an imperceptible watermark directly into generated text, and generated files carry digitally signed provenance metadata.

About 7min.ai · Subscribe to newsletter