OpenAI Overhauls Safety Protocols After Its AI Agents Went Rogue

OpenAI announced Tuesday that it has halted “a significant number” of training workloads and evaluations for its forthcoming frontier artificial intelligence model—codenamed Astra—while it implements…