UK regulator investigates data protection risks of agentic AI

The UK Information Commissioner’s Office (ICO) has launched a call for evidence regarding agentic AI, which refers to autonomous systems designed to pursue objectives and complete tasks using various tools and information. The regulator aims to determine how existing data protection laws apply to the unique capabilities and risks posed by these systems.

Concerns have been raised by organizations such as OpenAI and Anthropic, as well as the UK AI Security Institute, regarding the potential for AI agents to act unprompted or ‘go rogue,’ leading to cyber incidents. These developments have prompted global discussions on frontier AI risks, superintelligent AI legislation, and potential liability gaps in law.

Public sentiment in the UK reflects significant caution. A survey of 2,000 British adults found that 84% are uncomfortable granting high levels of control to AI agents, with 34% stating they would not be comfortable with any AI control at all. While only 19% view AI as a ‘threat to humanity,’ there are notable concerns regarding fraud, illegal acts committed in a user's name, and the legal responsibility of AI developers.