What happened
The Ada Lovelace Institute's policy briefing argues that Advanced AI Assistants have features that are 'more persuasive and engaging than older versions of chatbots' and can become a 'universal digital intermediary' between people and online information, ideas, products and services, creating autonomy risks that in aggregate threaten democracy and economic markets. It maps how the EU digital rulebook (DSA, DMA, AI Act, consumer law) partially covers these risks, then identifies gaps: unresolved scope questions on DMA CPS categories and the AI Act's GPAISR provisions, limited enforcement or case law, AI companies' partial disclosure of user and usage data, and the difficulty of proving causation to a legally actionable level. It makes detailed recommendations, including updating the concepts of the 'average' and 'vulnerable' consumer in the pending Digital Fairness Act, specifying and prohibiting Assistant-specific dark patterns, requiring paid-for advertising to be clearly disclosed and personalised advertising disabled by default, and asking regulators to interpret 'significant harm' in the AI Act's manipulation prohibition to include widespread, accumulative negative impacts.
Why it matters
This gives policy and EU-affairs leads a concrete, multi-venue roadmap for how the AI Act, DSA and the coming Digital Fairness Act should be applied to assistant-style agents — the next regulatory battleground for autonomous AI consumer products.
Action needed
Brief your EU compliance and product teams on the scope and evidence gaps it identifies so assistant products are designed defensively against upcoming guidance and enforcement.