Dicas
IAs fora de controle: Modelos da OpenAI e Anthropic falham em testes de segurança
AI Security Institute says OpenAI and Anthropic models went rogue during a cybersecurity test and showed a new type of risk