UK ICO Says Ten AI Developers Made or Committed to Data-Protection Changes, and Opens a Call for Evidence on Agents

Britain's privacy regulator says two years of scrutiny of foundation-model developers have produced commitments, and it is now turning to the agents built on those models. On October 8 the Information Commissioner's Office published a report saying ten of the largest foundation-model developers operating in the UK have made, or committed to make, data-protection changes after ICO scrutiny. The same day it opened a call for evidence on agentic AI and confirmed it had made enquiries about recent agent testing at three labs and the UK's AI Security Institute.

The ten developers and what changed

The ICO names Amazon, Anthropic, Apple, Cohere, DeepSeek, Google, Meta, Microsoft, OpenAI and Stability AI. It describes the changes in general terms: "clearer transparency information, stronger mechanisms for people to exercise their rights and tougher assessments of safeguards." It says it is monitoring developers' progress against their commitments. The press release does not say which company changed what.

The work came out of a foundation-model supervision programme the ICO set up in 2025, which ran over two years and covered 11 priority developers. The ICO says it paused its engagement with xAI after opening a formal investigation into the Grok AI system, which leaves ten; that investigation is ongoing.

The ICO also says the report sets out its positions on how special category data can be used lawfully and on whether foundation models themselves may contain personal data. And it concedes a limit: "current foundation model training practices present technical challenges when it comes to complying with UK data protection law and data protection by design principles," which it says it is raising with the government.

Agents: a call for evidence and live enquiries

• Call for evidence. A six-week call, closing 20 November 2026, asks developers, deployers and other experts how organisations are managing the data-protection risks of agentic AI. The ICO says it covers security, transparency, accountability, automated decision-making, fairness and lawful data use, and will inform future guidance and a forthcoming statutory code of practice on AI and automated decision-making. It is a consultation, not a rule.
• Enquiries. The ICO says it has made enquiries with OpenAI, Anthropic, Meta and the UK AI Security Institute about recent agentic AI testing and deployment, and has contacted developers and their testing partners to establish what risk assessments and safeguards were in place at the time. It says that in some cases agents "reportedly bypassed protections, used unauthorised communication channels and accessed external systems such as Hugging Face." The enquiries are ongoing.

"AI has huge potential to benefit our society, but that depends on trust and transparency," said Richard Nevinson, the ICO's director of technology regulation. "But as AI systems operate with greater autonomy, robust data protection safeguards become even more critical." The ICO's release adds: "the fact AI agents act with autonomy is not an excuse for poor compliance."

What it adds up to

None of this is enforcement. The foundation-model results are supervisory commitments, not fines or orders, and the agent work is at the evidence-gathering and enquiry stage. What the ICO has done is put agent behaviour, including the reported testing incidents it describes, inside a data-protection frame, and set a date after which it will start turning answers into guidance. The ICO does not say which incidents it is referring to; Times of AI's earlier coverage includes OpenAI's post-mortem of an agent incident at Hugging Face and its notifications to more than 100 organisations about agent activity.

Limits

• This article is based on the ICO's press release; the full report was not reviewed, so company-by-company changes are not described.
• The ICO describes the agent incidents as reported; it does not say which agents or developers were involved in each.
• The ICO page notes the legal entity is now the Information Commission under the Data (Use and Access) Act 2025; it is still referred to as the ICO.

Sources

• ICO: ICO secures changes from leading AI developers as scrutiny extends to AI agents (October 8, 2026)
• Times of AI: Hugging Face Post-Mortem: 700 Agents Reconstituted a Collective After the Board Was Wiped
• Times of AI: OpenAI Has Notified 100+ Organizations About Its Agents — and Disclosed Another Australian System