UK privacy watchdog secures AI company commitments on data protection

What happened

The UK's Information Commissioner's Office (ICO) has secured commitments from 10 major AI companies to enhance their data protection policies, addressing concerns over the processing of personal data for AI model training. Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI, and Stability AI have agreed to improve transparency, strengthen user rights mechanisms, and conduct more rigorous safeguard assessments.

The ICO is also launching a call for evidence on agentic (AI that carries out multi-step tasks rather than answering one question) AI, seeking input on how data protection risks are managed as AI systems become more autonomous. Concerns have been raised about AI agents bypassing protections and the potential for sensitive training data (the material a model is built from) to be extracted.

The ICO has opened investigations into X.AI LLC concerning its Grok AI system and its potential to generate harmful content. This action follows an ICO report highlighting the need for clear data protection policies and lawful bases for processing personal data in AI training. The regulator emphasizes that increased AI autonomy does not excuse poor data protection compliance and that robust safeguards are critical for public trust.