EU says AI agents covered

The European Commission has added a plain answer to its AI Act help desk: AI agents are covered by the law. (ai-act-service-desk.ec.europa.eu) The new entry appears in the Commission’s AI Act Service Desk frequently asked questions, which now includes the question, “How are AI agents addressed within the AI Act?” The page says the FAQ list is updated regularly based on stakeholder questions and AI Pact webinars. (ai-act-service-desk.ec.europa.eu) That matters because the European Union’s law does not create a separate legal bucket for “agents.” The Commission’s main AI Act page says the regulation applies a risk-based framework to artificial intelligence systems, with different rules for prohibited, high-risk, and other uses. (digital-strategy.ec.europa.eu) The timetable is already running. The Commission says the AI Act entered into force on August 1, 2024; bans on prohibited practices and artificial intelligence literacy duties started applying on February 2, 2025; obligations for general-purpose artificial intelligence models started on August 2, 2025; and the broader regime applies from August 2, 2026. (digital-strategy.ec.europa.eu, digital-strategy.ec.europa.eu) In practice, an “agent” is software that can take multiple steps toward a goal, such as calling tools, retrieving data, or triggering actions in other systems. The Commission’s public guidance already frames the law around whether something is an artificial intelligence system or a general-purpose model, not around the marketing label attached to it. (digital-strategy.ec.europa.eu, ai-act-service-desk.ec.europa.eu) The Commission has been building out that guidance layer for months. In December 2025, it said the European Artificial Intelligence Office was preparing practical guidelines to help companies apply the Act alongside other European Union rules. (digital-strategy.ec.europa.eu) Vendors are already selling around that gap between legal text and implementation. Techzine reported on April 14 that Cisco and Belgian artificial intelligence firm ML6 signed a memorandum of understanding to help organizations secure artificial intelligence applications and prepare for August 2026 compliance. (techzine.eu) ML6 said it joined Cisco’s AI Defense Design Partner Program as one of the first European partners. The company said Cisco’s product offers real-time protection for artificial intelligence applications across cloud, hybrid, and on-premises deployments. (ml6.eu) Cisco’s pitch is that agent-style systems need continuous testing before and during deployment. Techzine described AI Defense as an “algorithmic red team” and said Cisco added Zero Trust controls for AI agents and extra security for Model Context Protocol traffic in March. (techzine.eu) The Commission’s answer does not rewrite the Act, but it narrows a question companies had been asking in public: whether autonomous and semi-autonomous software sits outside the existing categories. The answer from Brussels is now shorter than the debate: agents are in. (ai-act-service-desk.ec.europa.eu, digital-strategy.ec.europa.eu)