首页 / ShortFling 博客 / Safety
本文目前仅提供英文版。阅读英文版 →
SafetyAI Star Mode Safety: Your Own Face, Consent, and Deepfake Laws
Where AI Star Mode stands
A feature that puts a face into a story is only acceptable if the face is yours and the whole thing is built around consent. AI Star Mode, the planned ShortFling feature that would cast you as the lead of a short-drama template, is not live: today there is only a waitlist, and none of our 12 drama series can be watched yet either.
So this post describes rules we are designing, not a shipped product. We are publishing them early because the law in the US, the EU, the UK and South Korea has spent the last two years moving in the same direction, and we would rather build to that direction than retrofit later.
The rules we are designing
Only your own face. AI Star Mode is meant for one person: the account holder. Not a partner, not an ex, not a coworker, not a celebrity, not anyone else who exists in real life.
Before any generation runs, we plan a live selfie check to confirm that the face being used belongs to the person holding the camera. A saved photo of someone else, or a still from a film, should not get through.
NSFW templates will be available only to users who are 18+, and only with their own likeness. That second condition is the one that matters most, because it is exactly where the laws below draw their hardest lines.
Results will be private by default, and you will be able to delete them. Our policy will be not to train public models on your face.
That is the full list. We are not promising anything beyond it here.
Why the own-face rule carries the weight
Most of the harm in AI face swapping comes from one situation: a real person's face used without their agreement. Asking a user to promise they have a friend's permission is easy to type and impossible to check. Limiting the feature to the face that passes a live check removes the third party entirely, which is why we treat it as the core rule rather than one setting among many.
United States: the TAKE IT DOWN Act
Signed on May 19, 2025, the TAKE IT DOWN Act makes it a federal crime to knowingly publish intimate images of a person without consent, and it explicitly covers "digital forgeries" made with AI. Its criminal provisions took effect immediately.
The platform duties became enforceable on May 19, 2026: sites hosting user content must offer a way to request removal and take down reported non-consensual intimate images within 48 hours. The FTC enforces that part.
European Union: AI Act Article 50
Article 50 of the EU AI Act applies from August 2, 2026. People who deploy AI to produce deepfakes must disclose that the content was artificially generated or manipulated; for evidently artistic, creative or fictional work, that disclosure can be made in a way that does not spoil the work. Providers must also mark AI output in a machine-readable way.
The 2026 Digital Omnibus changes left Article 50 on schedule, gave systems already on the market until December 2, 2026 for the machine-readable marking, and added a ban on AI systems that generate non-consensual intimate imagery, with a transition period to the same December date.
United Kingdom: sharing, then creating
Since January 31, 2024, sharing or threatening to share an intimate image of someone without consent has been an offence in England and Wales, inserted into the Sexual Offences Act 2003 by the Online Safety Act 2023. The wording covers images that "appear to show" a person, so deepfakes are included.
Section 138 of the Data (Use and Access) Act 2025 went further. From February 6, 2026, creating or requesting a non-consensual intimate deepfake of an adult is itself an offence, even if the image is never shared.
South Korea
On September 26, 2024, South Korea's National Assembly passed a bill making it a crime to possess or watch sexually explicit deepfakes, with penalties of up to three years in prison, and raising the maximum for creating them to seven years regardless of intent to distribute.
What the pattern says
Read together, these laws are not aimed at people using their own likeness in fiction. They target the use of someone else's face without consent, and the newer ones reach earlier in the chain: from sharing, to creating, to requesting, and in Korea to watching. A feature that only ever works on the verified account holder sits on the right side of every one of those lines by design, not by moderation after the fact.
This is a plain-language summary, not legal advice; rules differ by country and change, so check with a lawyer for your own situation.
If you want to try AI Star Mode when it opens, join the waitlist. Until then, our fiction library is live, with an 18+ NSFW section where every character is an adult and no real people appear.
FAQ
Is it legal to put your own face in an AI video?
The laws covered here, the TAKE IT DOWN Act, the EU AI Act, the UK offences and South Korea's 2024 amendment, focus on using someone else's likeness without consent. Using your own face is generally not their target, though rules vary by country and this is not legal advice.
Can I use a friend's face in AI Star Mode if they agree?
No. AI Star Mode is being designed for your own face only, with a planned live selfie check before generation. Celebrities and every other real person are excluded, with or without claimed permission.
When does AI Star Mode launch?
There is no launch date. The feature is waitlist-only for now; you can sign up at /features/ai-star.
Sources
- WilmerHale: The TAKE IT DOWN Act Goes Live (June 2026) (checked 2026-10)
- EU AI Act, Article 50: Transparency obligations (checked 2026-10)
- Gibson Dunn: EU AI Act Omnibus Agreement (checked 2026-10)
- legislation.gov.uk: Sexual Offences Act 2003, section 66B (checked 2026-10)
- DLA Piper: How the law is catching up with AI deepfakes (checked 2026-10)
- Doughty Street Chambers: The legal response to AI-generated sexual deepfakes (checked 2026-10)
- CBS News: South Korea set to criminalize possessing or watching sexually explicit deepfakes (Sept 2024) (checked 2026-10)




