Tech giants warn that window to defend against AI attacks is narrowing

Open letter signed by OpenAI and Anthropic among others signals doubt on ability to control AI models