Azerbaijan moves to criminalise non-consensual AI porn and mandate AI content labels
Azerbaijan's Parliament is weighing two draft laws that would set criminal penalties for creating sexually explicit deepfakes without consent and require AI-generated content to be clearly labeled. Committees discussed the bills on Thursday.
Under the criminal proposal, generating non-consensual sexually explicit content with AI could carry up to seven years in prison. A separate bill would make AI-content labeling mandatory, with fines of ₼80-₼150 ($50-$90) for unlabelled distribution.
Key provisions at a glance
- Criminal offence: AI-generated sexually explicit content without the depicted person's consent; penalty up to seven years' imprisonment.
- Labeling duty: AI content must be "clearly and conspicuously labeled"; failure to label triggers administrative fines (₼80-₼150).
Why this matters for legal teams
The criminal bill targets creators of non-consensual sexual content, but exposure could also arise for funders, repeat distributors, and facilitators depending on final wording. The labeling bill, if broad, may touch publishers, platforms, agencies, and corporate communications that use synthetic media.
Expect compliance questions around disclosure standards, recordkeeping, and proof of synthetic origin. If enforcement leans on intermediaries, duties could extend to notice handling, takedown, and provenance tooling.
Expert concerns: proof, process, and practicality
Information Communication Technologies expert Osman Gunduz flagged the surge in deepfakes and linked harms like reputational attacks and sexual blackmail, noting that many countries are developing legal tools to counter these risks. He warned that implementation details for media labeling are unclear: "How will the authenticity of AI-generated images be proven, what level of labeling will be required when using them, and how will their effectiveness be assessed when using additional tools?"
Gunduz also cautioned that penalties alone won't solve enforcement challenges and that adoption of AI tools across media and creative industries is accelerating; overly rigid rules could distort competition and create bureaucratic choke points.
Rule-of-law risks
Human rights lawyer Yalchin Imanov called this Azerbaijan's first legislative attempt to regulate AI-generated content. While he welcomed the move in principle, he questioned whether the laws would be applied fairly, pointing to past patterns of selective enforcement that shielded officials.
Open questions to watch in the draft text
- Definitions: What qualifies as "AI-generated" or "synthetic" content? Are partially edited works covered?
- Consent and intent: Is explicit consent required in writing? Is intent or knowledge (mens rea) an element for the criminal offence?
- Scope: Does liability attach to creation, distribution, or both? Are reposts and algorithmic amplification covered?
- Exemptions: Will there be carve-outs for news reporting, public interest, satire, or art? What about archival or academic use?
- Intermediaries: Are platforms, hosts, and advertisers given safe harbors conditioned on notice-and-takedown or due diligence?
- Labeling standard: What counts as "clear and conspicuous"? Placement, size, duration for video, and language requirements?
- Proof and forensics: How will authenticity be established? Will provenance metadata, hashes, or watermarking be recognized?
- Extraterritoriality: How are cross-border creators or platforms addressed? What cooperation mechanisms will be used?
- Sanctions ladder: Are there escalating penalties for repeat violations, and are there compliance credits for good-faith efforts?
Comparative context
Transparency obligations for synthetic media are emerging in multiple jurisdictions. The EU's AI Act, for example, introduces disclosure duties for certain AI-generated or manipulated content, including deepfakes, to protect users from deception. See the European Commission's overview of the initiative for broader context: EU AI Act.
Immediate steps for counsel and compliance
- Map where your organization creates, edits, or distributes AI-generated media (marketing, newsrooms, HR, product, vendor content).
- Draft a labeling policy that covers text, image, audio, and video, with placement and persistence rules; maintain versioned templates.
- Adopt provenance practices (asset logs, original file retention, cryptographic hashes; consider interoperable metadata standards).
- Update contracts with agencies, freelancers, and vendors: disclosure warranties, consent representations, indemnities, takedown SLAs.
- Refresh content moderation and incident response for deepfake reports; define timelines, evidence thresholds, and escalation paths.
- Train staff on consent requirements for synthetic sexual content and on prohibited uses; document completion.
- Plan for audits: keep evidence of labeling, consent records, and decisions tied to complaints and removals.
What's next
Watch for committee markups that define scope, enforcement bodies, and technical standards. The practical impact will turn on how "AI-generated" is defined, how proof works in court, and whether intermediaries get workable safe harbors.
For a deeper skill build on governance, liability, and enforcement around deepfakes and AI content, see AI for Legal.
Your membership also unlocks:
