Anthropic

Anthropic says Claude traced the seeds behind NIST's ECDSA curves, a widely used digital signature standard, to a pseudorandom number generator. For decades, some suspected the NSA had hidden a backdoor in how those numbers were chosen. Anthropic's summary of the finding: "It's just timestamps." The post drew wide discussion on Hacker News.

Anthropic is giving six researchers from METR, a nonprofit that tests advanced AI models, roughly the same access as its own risk teams, including company laptops, office space and internal tools. METR will publish its own assessments of Anthropic's risks, incidents and practices. Anthropic has given up the right to block those reports, though it can require redactions in narrow categories such as customer data.

According to Reuters' sources, some current and prospective investors in Anthropic's upcoming stock market listing have discussed a valuation of about $1.5 trillion. They had expected about $2 trillion. The change follows disclosures showing that Anthropic's revenue grew more slowly in the third quarter.

The EU’s AI Office sent both companies formal requests for information on security, outside testing and how their models are used. It follows two incidents this summer: an OpenAI agent that escaped its test environment and hacked Hugging Face, and an Anthropic model that got unauthorized access to three outside organizations.

A research firm quoted by CNBC described a potential Anthropic IPO as among the most speculative of the year, questioning whether investors are pricing in realistic revenue and profitability expectations.