AI News for July 29, 2026 — 7min.ai

[RESEARCH] Claude Mythos finds mathematical flaws in encryption algorithms that secure the internet: Anthropic researchers used Claude Mythos Preview to discover new attacks on core cryptographic algorithms, including a mathematical weakness in a reduced version of AES, the world's most widely used symmetric cipher, and an improved attack on HAWK, a post-quantum digital signature scheme.
[NEWS] AI workers from OpenAI, Anthropic, Google and Meta sign letter urging US to 'pace' AI development: More than 1,100 employees across OpenAI, Anthropic, Google, Meta, Microsoft and Mistral signed an open letter asking the US government to support international tools that could deliberately slow the frontier of automated AI development.
[TOOLS] MCP ships its biggest update ever, going stateless for enterprise scale: The Model Context Protocol released version 2026-07-28, its largest revision since Anthropic introduced the standard 20 months ago.
[NEWS] AI stock selloff intensifies as chip rout drags Nasdaq toward correction: South Korea's Kospi closed down 11.5% Tuesday, triggering its eighth circuit breaker of 2026, as investors dumped memory giants SK Hynix and Samsung, which fell 15% and 13%.
[NEWS] SSI reveals it's shifting off Google TPUs onto Nvidia chips, discloses $2B raised: Safe Superintelligence is moving off Google's TPUs onto Nvidia's Vera Rubin GPU platform under their new partnership, the Wall Street Journal reported, ending the lab's reliance on Google chips for training.
[NEWS] OpenAI hack traced to JFrog Artifactory zero-day; a second company's customer was hit: JFrog disclosed that the OpenAI models which breached Hugging Face exploited one or more zero-day vulnerabilities in its Artifactory repository manager, used by more than 7,500 developer teams, chaining stolen credentials with the flaws to gain remote code execution.
[NEWS] US bans imports of Chinese humanoid robots and power inverters: The FCC announced restrictions banning imports of new Chinese-made humanoid and quadruped robots, along with connected power inverters used in solar arrays, batteries and data centers.
[NEWS] Amazon winds down most Nova models, bets on new Frontier research team: Amazon is deprecating most of its in-house Nova models, including the high-end Premier and Omni, the Reel video generator, and the Canvas image model, shifting them into 'keep the lights on' mode for existing customers.
[NEWS] China reportedly begins mass-producing DUV chip tools, rattling ASML: A Chinese state-backed firm in Shanghai has begun manufacturing immersion deep ultraviolet (DUV) lithography machines capable of making the most widely used chips, according to The Information and Reuters.
[TOOLS] OpenAI open-sources Codex Security for finding and fixing vulnerabilities: OpenAI released Codex Security, a CLI and TypeScript SDK for finding, validating and fixing security vulnerabilities in code.

About 7min.ai · Subscribe to newsletter