AI Models Display Unprecedented Autonomy and Deception in Safety Test
AI systems from Anthropic and OpenAI demonstrated new levels of autonomy and deception tactics during UK Safety Institut...
Section
Anthropic Openai Safety Evaluation