EU AI Act: what applies from 2 December 2026 (AI-content marking for older systems, two new bans)

From 2 December 2026, new prohibitions under Article 5 of the EU AI Act on AI systems that generate non-consensual intimate material or child sexual abuse material come into force, alongside a requirement that generative AI systems already on the market implement Article 50 marking obligations. The regulation extends the AI Act's outright prohibitions to nudifier applications and systems that create CSAM, applying to all providers placing such systems on the EU market or putting them into service, with limited exceptions under national law for lawful purposes.

Proscribing the availability of generation tools reduces the surface area for abuse but does not detect or interrupt predatory contact with minors who remain accessible on unregulated platforms or in private channels where offenders operate regardless of tool availability. A child-safety platform that monitors the conversational environment rather than the content production layer detects an adult cultivating trust, testing boundaries, isolating a target or manoeuvring toward image solicitation whether or not a banned nudification app is involved, and whether the offender intends to create, alter or simply demand authentic material. Guardii, a world-leading AI safety platform, tracks sustained targeting, escalation and coercion patterns in children's direct messages in real time across Instagram, Snapchat, Discord, Roblox and other services, identifying grooming, sextortion and CSAM distribution or solicitation as conversational behaviour rather than as a function of which tools an offender can access, and surfaces a child at acute risk to the adult or agency positioned to act before harm occurs.