Creating sexually explicit deepfake images to be made offence in UK
Definition and Legal Scope of “Deepfake”
- Several commenters note the UK announcement lacks a clear definition of “deepfake,” raising concerns that anything from Photoshop edits to AI outpainting might qualify.
- Comparisons are made to UK child sexual abuse material laws, where even drawings can be illegal, suggesting this law may follow an “end result” rather than “production process” logic.
- Others distinguish AI-generated imagery (modeled likeness) from simple face superimposition, and question why separate legislation is needed given existing laws on intimate images and harassment.
Creation vs Distribution (“Thoughtcrime” Debate)
- A major fault line: whether merely creating explicit deepfakes should be criminal if they are never shared.
- One side argues creation alone is victimless; harm arises only when images are distributed, believed, and used for harassment, blackmail, or reputational damage.
- The other side stresses that:
- People cannot guarantee private data won’t leak.
- Deepfakes are inherently degrading and dehumanizing, even if initially private.
- Law should treat “creation with potential to leak” as culpable.
- Some see criminalizing private, unshared files as edging into “thoughtcrime.”
Harm, Consent, and Scale
- Many emphasize severe harms: sexual harassment, long-term trauma, bullying (including of minors), and potential suicides; deepfakes are likened to “visual libel” with high believability.
- Others compare deepfakes to fan-art, erotic fiction, or sketches inspired by real people and ask why these remain legal while deepfakes would not.
- The reduced cost and speed of AI generation is viewed as transforming a niche problem into a mass-scale threat.
Free Expression, Art, and Defamation
- Debate over whether people have a right to produce sexualized depictions of real others without consent, as with caricatures or unflattering paintings.
- Some argue explicit depictions cross a line, especially given the potential to pass as real; others think existing defamation, fraud, and harassment laws should apply instead of broad bans on creation.
Enforcement Practicality and Gray Areas
- Skeptics doubt enforceability given anonymous, mass AI generation and global adult sites.
- Numerous edge cases are raised: low-quality or obviously fantastical deepfakes, clearly labeled synthetic images, altered or partially censored images, accidental resemblance, and similarity thresholds—highlighting likely ambiguity and litigation.