Source: JD Supra

Generative AI’s growing role as an enabling technology has prompted regulators to focus more on user-facing conversational tools—labelled variously as “AI agents”, “AI chatbots” and “virtual assistants”—because these tools shape access to information and services and raise competition and consumer-protection issues. The piece distinguishes chatbots (conversational responses/limited tasks), virtual assistants (intermediary gateways to other services/devices), and true AI agents (autonomous systems that perceive, decide and act across tools and data) and stresses that legal analysis should depend on actual functionality rather than labels. Key compliance themes include transparency about system limits, testing and human oversight where autonomy affects commercial decisions, planning for interoperability where agents access third-party platforms, and assessing product design for risks of self‑preferencing or lock-in.

EU digital legislation only partially captures these technologies: the DMA includes “virtual assistants” as a Core Platform Service but does not clearly categorise AI agents; the AI Act applies broadly with transparency obligations and possible high‑risk classification; and the DSA’s applicability depends on whether AI agents are “intermediary services”, a point under assessment by the European Commission and already contested between the Italian Communications Authority and the Italian Competition Authority. Recent enforcement trends highlighted include platform efforts to restrict agents’ access to third‑party services (“platformisation of AI agents”) and consumer risks from hallucinations or manipulation by conversational AI. The EC’s DMA review and decisions (e.g., Article 6(11) on search data sharing) indicate it may treat some AI services as falling within DMA obligations and consider extending the DMA’s scope or adding AI services as new CPSs, while also relying on existing antitrust tools for conduct outside the DMA.