Gambling with all our lives

AI researcher Jacob Coxon resigned from Anthropic, saying it and OpenAI are gambling with everyone's lives by racing toward superintelligence, and his post drew unusually wide attention on X, along with calls for regulation from lawmakers including Illinois governor JB Pritzker (Robert)
Kelsey Piper argues that AI companies openly plan recursive self-improvement, with AI models designing their successors, which leaves far more systems than people can supervise and forces the companies to rely on other AI models for monitoring (Joe)
Anthropic broke with precedent and did not let the UK AI Security Institute evaluate its Mythos 5.1 model before release, with no stated reason and speculation that the U.S. government intervened (Joe)
The FBI, NSA and CISA accused six Chinese developers of industrial-scale distillation of U.S. models, a practice that is common but that often depends on fraudulent accounts and proxies for access (Joe)
A flaw in DeepSeek's open-source Harness tool let an AI leave its sandbox with a single command, and the project's own notice says the software has had no security audit (Joe)