Nvidia's AI Financing Guarantees Face First Test as Neocloud Bonds Hit Distress
Nvidia has pledged more than $500 billion in off-balance-sheet support to the AI buildout, including up to $105 billion behind a single OpenAI data center. Now the bonds of the neoclouds it backs, CoreWeave, Nebius and IREN, are trading at distressed levels, in the first real test of the chipmaker's turn to financier. Jensen Huang says the problem is a shortage of capital, not demand; skeptics say it is circular financing propping up lofty expectations.
Read full story →AI-Discovered PHP Flaw Knocks Government Websites Offline Across Europe
A flaw found with an OpenAI model in PHP, the language behind much of the web, forced government sites offline this week, with some operators pulling their own sites down to stay safe. Researchers say AI has doubled the number of vulnerabilities discovered year on year, and Cloudflare's CEO warns of a couple of years in which "every single week it's going to feel like the sky is falling."
Read full story →Moonshot AI Unveils Kimi K3.1 Pro and Deliberately Cuts Its Cyberattack Skills
China's Moonshot AI says it stripped cyber-offensive data from its new open-weight model before release, lowering its internal hacking benchmark score from 90.5% to 81% in exchange for reduced misuse risk. Kimi K3.1 Pro otherwise targets games, 3D design and financial research, with full weights due November 4. For an open model, the safeguard is close to irreversible once the weights circulate.
Read full story →UK Commits £600 Million to AI Security, With Biggest Share for Biosecurity
Britain set out about £600 million ($800 million) on Saturday to defend against health and security threats from AI, led by a new biosecurity lab, with OpenAI and Anthropic offering up to £200 million in model credits. The package adds £130 million for critical-infrastructure cyber defence, but lands amid a government warning that the UK risks becoming a "high-value AI customer" rather than a builder.
Read full story →