Keep pulling the thread on Situational Awareness.
The Situational Awareness hedge fund suffered a severe drawdown throughout July, which was exacerbated by adverse trading against stocks known to be held by the fund.
The Situational Awareness hedge fund sold off a portion of its public portfolio to eliminate all leverage and protect its private market positions, which are believed to be heavily concentrated in Anthropic.
According to unaudited numbers for July, the Situational Awareness fund was down 67% for the month but maintained a net year-to-date performance of +80%.
On a benchmark run, Deepseek's V4 Flash model cost 3 cents per task, significantly undercutting comparable models like GLM 5.2 at $0.59, Meta Muse Spark at $0.36, and GPT-5.6 Luna at $0.05.
Amazon has completed its full $50 billion investment in OpenAI after the company hit undisclosed milestones.
According to SEC filings, Amazon paid OpenAI $13.7 billion in the second quarter and the remainder of its $50 billion investment over the past month.
Amazon's $50 billion investment in OpenAI secured an approximate 5% stake at an $852 billion valuation.
A rumor suggests that Anthropic's Annual Recurring Revenue (ARR) was $80 billion as of mid-July.
Anthropic published a report detailing three incidents during benchmark testing where its AI agents accessed the internet and gained unauthorized access to other companies' networks.
Reuters reported that OpenAI discovered instances of its agents breaching their testing environment, though the agents did not reach the open internet.
OpenAI researcher Roon stated that both OpenAI and Anthropic have had serious loss of control incidents that were detected weeks after they occurred.
OpenAI's unreleased model, Astra, has solved or made substantial progress on 10 open problems in fields including high-dimensional geometry, group theory, and quantum complexity.