AI Act: two new prohibitions become applicable
What applies
The Digital Omnibus expanded the AI Act's list of prohibited practices. Two new prohibitions are added: AI systems that generate or manipulate realistic intimate depictions of an identifiable person without their explicit consent, and AI systems that generate or manipulate depictions of child sexual abuse (Art. 5(1), first subparagraph, points (ba) and (bb) AI Act, as amended by Regulation (EU) 2026/1744).
They only become applicable on 2 December 2026 — that's stated explicitly in Art. 113(a). Until then, this is upcoming law, not yet in force. We're covering it now anyway, because four months' lead time is exactly the point of a transition period.
For companies that use AI, the scope is drawn narrowly: under Art. 5(1a)(b), the prohibition only applies to a deployer when the system is used for the purpose of such generation. A general-purpose image generator isn't prohibited just because it's technically capable of it.
Who it affects
Directly: providers and developers of generative image and video systems. For every other company, it's a matter for the AI usage policy and tool selection.
Example (fictional): The tool without safeguards
An advertising agency evaluates three image generators for campaign visuals. Two document technical safeguards against misuse; the third — the cheapest — says nothing about it and markets itself with “no restrictions.”
The agency picks one of the other two and records the rationale in its AI usage policy. Six months later, when a major client asks about tool selection during a supplier audit, the answer is a document instead of a phone call.
The risks, plainly stated
Highest fine bracket in the Regulation
Violations of the Art. 5 prohibitions carry fines of up to €35 million or 7% of worldwide annual turnover (Art. 99(3) AI Act). For SMEs and start-ups, the lower of the two figures applies (Art. 99(6) AI Act).
Tools already in the building
Prohibitions don't only hit new purchases. If you don't know which image generators are already in use, you can't answer this question in December.
Legal territory beyond the AI Act
The content described is criminally relevant independent of the AI Act. The new prohibitions sit alongside that law — they don't replace it.
Your next steps
- 01Check the AI inventory: which generative image and video tools are actually in use?
- 02Record in the AI usage policy what these tools may not be used to generate
- 03Make documented provider safeguards a condition of tool selection
In Compliverse
The compass turns exactly these steps into measures in your plan automatically — with deadlines, owners, and evidence.
Plain-language product content, not legal advice · Case studies are fictional · Fine amounts as stated in the legal act (“up to”)