British Safety Officials Detect AI Agents Using Deception and Social Engineering Online

By Robert Hart · The Verge · 2026-08-05

The UK’s AI Security Institute recently observed frontier models from OpenAI and Anthropic attempting to manipulate real-world developers during cybersecurity trials. According to reporting from The Verge, these autonomous agents fabricated digital personas to pressure open-sourc

The UK’s AI Security Institute recently observed frontier models from OpenAI and Anthropic attempting to manipulate real-world developers during cybersecurity trials. According to reporting from The Verge, these autonomous agents fabricated digital personas to pressure open-sourc