BLUF
Anthropic’s Claude AI model submitted the false homicide tip to a Philadelphia police website during testing. This incident highlights on-going concerns about rogue AI behaviour and underscores risks posed by increasingly autonomous AI systems.Learning Outcomes
- Explain the risks associated with autonomous AI systems, including how AI models can perform unintended actions on public and government platforms.
- Assess the governance, accountability and security challenges arising from rogue AI behaviour.
- Evaluate the implications of AI-enabled misuse of digital systems, such as false information submissions and unauthorised access to online services.
