September 4 review: Anthropic described an eleven-day effort to formalize Fermat’s Last Theorem. Reports of Nscale financing and Anthropic’s IPO timetable showed continued demand for capital. Chrome patching and proposed US-China talks put security on the same agenda.
AI / Models65d ago
Claude formalizes a proof of Fermat’s Last Theorem
Anthropic said Claude spent 11 days translating a proof of Fermat’s Last Theorem into Lean so a computer could check it. The advance concerns formal verification of an established theorem, rather than the first discovery of its mathematical proof.
Google released a Chrome update addressing an actively exploited V8 vulnerability and other flaws, according to BleepingComputer. The report identified the renderer process as the affected context; it did not say every user had been compromised.
Reuters reported that US and Chinese officials were preparing to discuss AI safety in mid-September. The report described planned talks and named Treasury Secretary Scott Bessent as the expected US lead; no agreement had been announced.
Report outlines Anthropic’s possible IPO timetable
Reuters reported that Anthropic could publish its prospectus in late September and complete a listing before the November US midterms. The schedule was attributed to sources and remained a plan rather than a completed public offering.
Reports said AI cloud company Nscale was negotiating up to $3.5 billion in financing ahead of a planned IPO. Bloomberg’s account included a possible Nvidia investment, but the financing had not been presented as closed.
Bloomberg reported that AI data startup Micro1 offered $12.5 million for Spirit Airlines’ records. An existing $10 million agreement with Google complicated the proposed deal, illustrating competition for company data that can support AI training.
Nvidia said it planned to extend DLSS 5 to RTX 40-series graphics cards after further tuning RTX 50-series performance, The Verge reported. This was a support plan, without a confirmed delivery date in the report.
California investigates OpenAI’s Hugging Face incident
Politico reported that California Attorney General Rob Bonta was investigating OpenAI over the July Hugging Face incident. The inquiry added to scrutiny from other states; an investigation itself is not a finding of liability.