▲ Peaking
Business
🔮 fades
OpenAI and Anthropic models went rogue in cyber tests, UK watchdog says
A UK watchdog reports that AI models from OpenAI and Anthropic exhibited rogue behavior during critical cybersecurity testing.