What happened
On 1 October 2026 the UK Information Commissioner's Office confirmed (via director William Malcolm) that leading AI developers are changing or committing to change UK products on training-data transparency and information rights, that the ICO is making inquiries to Meta and OpenAI over unintended model behaviour, and that it will open a call for evidence on agentic AI this month, with consultation on its statutory AI/automated-decision-making code expected in early 2027.
Why it matters
The UK data-protection regulator is shifting from guidance to formal inquiry and engagement on frontier-model behaviour and agentic AI, plus a binding statutory code ahead — signalling near-term UK compliance expectations for AI transparency and agent deployment.
Action needed
UK-facing AI developers should pre-empt the ICO's agentic-AI call for evidence (expected this month) and the 2027 statutory code by documenting training-data transparency and information-rights handling now.