Privacy / AI Governance / Regulation
ICO says LLMs and agentic AI add no new data protection rules for online advertising but change compliance
The UK regulator's 'AdTech 3.0' report says advertisers stay accountable when AI agents run campaigns, and that inferred sensitive traits used for targeting can be special category data.

The Information Commissioner’s Office published a report on 13 November 2026, titled ‘AdTech 3.0’, on how large language models and agentic AI interact with PECR, the UK GDPR and the Data Protection Act 2018 in online advertising. It covers ads beside chatbot responses, agent protocols such as AdCP and ARTF, walled gardens, retail media and connected TV.
The ICO’s central finding is that these technologies do not create new data protection requirements, but may change how organisations apply them and demonstrate compliance. Accountability cannot be passed to an agentic system. According to Hunton’s summary, the report treats uncertain inferences of sensitive traits used for targeting as special category data needing explicit consent.
The ICO says the findings will inform its forthcoming work on agentic AI and automated decision-making, including the AI and ADM code of practice.
Why it matters
Teams deploying AI agents in marketing should be able to show they understand each agent's objectives, the personal data categories it can reach, the actions it may take and where humans can intervene, since, according to Hunton's summary, the ICO says meaningful oversight requires more than reviewing campaign results after the fact. Inference pipelines that derive health or other sensitive traits from chatbot inputs should be identified, as the ICO says such inferences may qualify as special category data where used to infer or act on a sensitive characteristic, for example in ad targeting. Vendor due diligence should seek evidence of data sources, processing flows and agent authority rather than relying on labels such as 'privacy-preserving', and automated workflows should be able to communicate and honour consent withdrawals across participants.
Status: Published. applies existing law; creates no new legal requirements
Sources
Was this useful?
Thanks. What was the problem?
Thank you — this helps us choose what to cover. If you spotted an error, please tell us what it is.
Prepared with AI assistance and reviewed by the editor before publication. Spotted an error? Tell us — see our editorial policy and methodology.
Governance areas
Institutions
Who should care
- DPO
- AI governance lead
- Legal & compliance
- Data stewards