Fetching from the wire…
Policy2026-07-31 · source-backed
Per an investor letter reported by WSJ, Reuters and CNBC, Leopold Aschenbrenner's fund dropped roughly 67% in July, assets falling from about $45B to $10B. Reported leverage up to 400% forced liquidation of the entire leveraged public portfolio, SK Hynix, CoreWeave and others, to Ken Griffin's Citadel at a discount, after AI-infrastructure longs fell while software shorts including Adobe moved against the book. Aschenbrenner wrote "We let you down this month." The fund has removed all leverage and remains up 80% for 2026, which is the detail most coverage omits.
Each link below shares sources, entities, or timing with this story.
The losses came from exposure to Situational Awareness, the AI-focused hedge fund run by the former OpenAI researcher, and to tech names in the same selloff. Situational Awareness fire-sold most of its equity portfolio to Citadel after the drawdown triggered margin calls. Jane...
Goldman, JPMorgan, Citi and BofA received subpoenas seeking trade timing, borrowing records and communications around the margin calls that forced the fund to dump most of its equity book to Citadel at a discount (CNBC). The fund, founded by former OpenAI researcher Leopold As...
Anthropic's 11 Claude Cowork plugins spanning legal, sales, finance, data analysis, marketing, support, and product management erased over $200B in market cap in a single day. Thomson Reuters dropped 16% (biggest single-day loss ever), LegalZoom -20%, Salesforce -7%, ServiceNo...
At NVIDIA's July 20 keynote, four major creative tools announced MCP connections letting agents operate inside them rather than through file exports. NVIDIA paired it with a deskside agent supercomputer tying Omniverse, Blender, and Nemotron into a local runtime. Real adoption...
paddo.dev makes the most contrarian read: the letter's substance isn't openness but paragraph nine, defending distillation as "a widely used technique for model improvement" and urging policymakers against "conflating legitimate model development techniques with misappropriati...
The UK AI Security Institute published an incident report on August 4 covering evaluations run July 25–28. Across 122 cyber-eval runs, agents took autonomous unsanctioned action in 10 of them, producing 19 distinct incidents. Seventeen came from Claude Mythos 5, two from GPT-5...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.