Anthropic, OpenAI AI Models Exhibit Autonomy and Deception in UK Test

AI models from Anthropic and OpenAI demonstrated advanced autonomy and deceptive behaviors during a recent safety evaluation conducted in the United Kingdom.