Skip to main content
High reported severity Investigation Opened

Manhattan DA says 12 seized websites sold AI-generated sexual deepfakes of about 1,200 people; the depictions' dates and an exact count are not established

On 14 September 2026 the Manhattan District Attorney's Office announced that it had seized, under a court order, 12 domain names of websites it says were used to disseminate, publish and sell non-consensual 'deepfake' videos, and said its investigations were ongoing. The office alleges that people under investigation used AI image and video creation tools to turn existing photos and videos of approximately 1,200 real people, overwhelmingly women and mostly public-facing figures, into sexual images and videos that were then sold. The 1,200 figure is the office's estimate and is not a documented count. WIRED reported that it had observed seizure notices on several of the sites in the days before the announcement and quoted researchers who described the videos as pornographic films manipulated to show the targets' faces, many apparently reposted from the defunct MrDeepFakes site. When the depictions were made and published is not established; one site claimed to have been online since 2018.

AI system
Unidentified image and video tool
Occurred
Event date unknown
Reported
14 September 2026
Event location
Unknown
What the AI did
Depicted or impersonated the person
Reported harm
Exploitation or Abuse
Whose AI use
Someone else’s AI use
Setting
Privacy · Justice
Evidence
AI involvement reported · Causal attribution alleged · 2 sources
5 claims: 1 documented, 4 reported. 6 open questions
People reported harmed
Not reliably quantifiable from the sources

AI system as recorded: AI image and video creation tools, not identified, which the Manhattan DA alleges were used to create the non-consensual sexual imagery

What Happened

The Manhattan District Attorney's press release of 14 September 2026 announces the seizure of 12 domain names of websites it describes as used for unlawfully disseminating, publishing and selling non-consensual celebrity deepfake videos. It says the sites were seized pursuant to a court order, that the office will monitor them to stop them reappearing under other domains, and that investigations into these and similar operations are ongoing. It does not name the sites, identify operators or report arrests. It asks survivors to contact the office's Cyber Crime Bureau.

The release alleges that individuals under investigation used AI image and video creation tools to turn pre-existing photos and videos of approximately 1,200 real people into what appear to be hyper-realistic images and videos of those people engaging in sexual conduct, and then used the websites to disseminate, publish and sell them. It describes the approximately 1,200 victims as overwhelmingly women and primarily public-facing individuals, including actors, politicians, athletes, musicians, social justice advocates and social media influencers. District Attorney Bragg's statement says such violations follow victims into their careers and personal lives and take an immense toll on emotional and mental wellbeing. These are the prosecuting office's allegations and general statements; the release does not describe individual outcomes, ages or a counting method, and the figure is an approximation rather than a lower bound.

WIRED reported the announcement the same day. It attributes the 1,200 figure to the office. Separately, it says it had independently observed several of the sites' homepages replaced by 'THIS DOMAIN HAS BEEN SEIZED' notices late the previous week, with the notices citing a seizure warrant issued by a New York State Supreme Court; the warrant itself was not inspected. WIRED says archived versions of the sites showed some videos watched tens of thousands of times and one site claiming to have been online since 2018. A researcher at the American Sunlight Project told WIRED that most videos were pornographic films manipulated to show the targets' faces, that many appeared to be reposts of content first shared on the MrDeepFakes website, and that at least two of the seized sites gave elected officials profiles with their names and positions. WIRED's figures for politicians combine people mentioned with people who appeared and are not used here as a count of depicted people. An Institute for Strategic Dialogue researcher called the takedowns a significant win and said the seized sites did not appear reachable through VPNs.

The seizure announcement is the response to the reported abuse, not the abuse itself. The inspected sources do not establish when the depictions were created or published, where the people depicted or the operators are, which tools were used, or any adjudicated finding against the unidentified operators.

Reported harm

The Manhattan DA alleges that about 1,200 real people, overwhelmingly women, were depicted without consent in AI-generated sexual images and videos published and sold on 12 websites it seized. The DA says such violations take an immense toll on victims' wellbeing and careers; no individual outcome, age or exact count is established.

How Harm Occurred

Alleged: AI image and video tools used by others to turn real people's photos and videos into sexual imagery that was published and sold without their consent (the DA's allegation; researchers quoted by WIRED describe face-swapped pornographic videos on the sites).

Outcome

Ongoing

On 14 September 2026 the DA's office announced the seizure of the 12 domains pursuant to a court order and said its investigations into the websites and similar operations were ongoing; it did not name the sites, identify operators or report arrests, and said it would monitor for the sites reappearing under other domains. WIRED reported that the seizure notices cite a New York State Supreme Court seizure warrant, which was not inspected.

What remains unknown

  • An exact or minimum number of depicted people and the office's counting method; the 1,200 figure is an estimate.
  • When the depictions were created and published; one site claimed to have been online since 2018.
  • Which AI tools were used and where the generation, hosting and the depicted people were located.
  • The identities of the operators and uploaders; no arrests or charges are reported.
  • Ages of the depicted people and any individual outcome; the DA speaks generally about the toll on victims.
  • The seizure warrant and its date; WIRED reports it from the sites' notices and it was not inspected.

What the evidence supports

AI involvement: reported. The DA's release states that individuals under investigation used AI image and video creation tools to turn pre-existing photos and videos of approximately 1,200 real people into hyper-realistic sexual images and videos that were then disseminated, published and sold on the 12 seized sites; the non-consensual depiction and its sale are the reported harm. A researcher quoted by WIRED describes the videos on the sites as pornographic films manipulated to show the targets' faces, and WIRED reports the sites advertised the material as deepfakes. The specific tools, the original files and the generated material were not inspected, and the account of the AI mechanism rests on the prosecuting office.

5 claims: 1 documented, 4 reported. What the statuses mean

Documented On 14 September 2026 the Manhattan District Attorney's Office announced that it had seized 12 domain names of websites it says were used for unlawfully disseminating, publishing and selling non-consensual deepfake videos, that the seizure was pursuant to a court order, and that its investigations were ongoing.

Causal attribution. The official release directly establishes the date and content of the office's announcement. It does not establish the truth of the underlying allegations, identify the sites or operators, or report any charge.

  • manhattanda.org(opens in new tab) supports · English
    'September 14, 2026'; 'today announced the Office’s seizure of 12 domain names of illegal websites used for unlawfully disseminating, publishing, and selling non-consensual celebrity “deepfake” videos'; 'we have seized these websites pursuant to a court order'; 'Our investigations into these websites and similar operations are ongoing'.
  • wired.com(opens in new tab) context · English
    'On Monday, the Manhattan District Attorney’s Office announced the seizure of 12 domains'.
Reported The DA's office alleges that individuals under investigation used AI image and video creation tools to turn pre-existing photos and videos of approximately 1,200 real people into sexual images and videos that were then published and sold on the sites, and describes the victims as overwhelmingly women and primarily public-facing figures (actors, politicians, athletes, musicians, social justice advocates and social media influencers).

Causal attribution. The AI mechanism, the approximate figure and the description of the victims are the prosecuting office's allegations; WIRED attributes the same figure to the office, so the two passages share one origin. The figure is an estimate, not a documented minimum, and no counting method, ages or individual outcomes are given.

  • manhattanda.org(opens in new tab) supports · English
    'Individuals under investigation allegedly used artificial-intelligence (“AI”) image and video creation tools to turn pre-existing photos and videos of approximately 1,200 real people into what appear to be hyper-realistic images and videos of those individuals engaging in sexual conduct'; 'They then used these websites to disseminate, publish, and sell these videos and other similar non-consensual intimate imagery'; 'The approximately 1,200 victims of these websites and the content creators who use these websites to sell their NCII videos are overwhelmingly women and primarily public-facing individuals, including actors, politicians, athletes, musicians, social justice advocates, and social media influencers'.
  • wired.com(opens in new tab) supports · English
    'Around 1,200 people—overwhelmingly women—were depicted in nonconsensual sexual images and videos on the websites, the office says'.
Reported WIRED says it independently observed several of the sites' homepages replaced by seizure notices late in the week before the announcement, with the notices citing a seizure warrant issued by a New York State Supreme Court, and that archived versions of the sites showed some videos watched tens of thousands of times and one site claiming to have been online since 2018.

Causal attribution. The reporter's own observations of the notices and archived pages. The warrant was not inspected, and a site's claim about its own age does not establish when any depiction was made.

  • wired.com(opens in new tab) supports · English
    'WIRED independently observed several of the sites’ homepages being replaced by takedown notices'; 'late last week, with the sites now saying they have been taken over following a seizure warrant issued by a New York State Supreme Court'; 'Archived versions of the websites show some videos had been watched tens of thousands of times, with one of the websites claiming to have been online since 2018'.
Reported A researcher at the American Sunlight Project told WIRED that most videos on the sites were pornographic films manipulated to show the targets' faces, that many appeared to be reposts of content first shared on the MrDeepFakes website, and that at least two of the seized sites gave elected officials profiles with their names and positions.

Causal attribution. Researcher observations reported by WIRED. The article's politician figures combine people mentioned with people who appeared and are internally inconsistent, so they are not used as a tally of depicted people.

  • wired.com(opens in new tab) supports · English
    'Most videos took the shape of actual pornographic movies, manipulated to instead show targets’ faces instead of the original actors'; 'many videos appeared to be reposts of content that had originally be shared on the MrDeepFakes website'; 'On at least two of the seized websites, elected officials were given profiles containing their name and position, along with a grid of manipulated videos hosting their likeness'.
Reported District Attorney Bragg said the violations of privacy follow victims into their careers and personal lives and take an immense toll on emotional and mental wellbeing.

Causal attribution. A general statement by the prosecutor about this kind of harm; it does not establish any individual's outcome.

  • manhattanda.org(opens in new tab) supports · English
    'These horrific violations of privacy follow victims into their careers and personal lives and take an immense toll on emotional and mental wellbeing'.
  • wired.com(opens in new tab) context · English
    'district attorney Alvin Bragg said in a statement'.

Sources

2 sources inspected. Sources that repeat one account do not corroborate each other.

How the sources were read, and where the events happened

Read in English on 2026-10-07 (live HTTP 200; also read 2026-09-17 and 2026-10-06). Press release dated September 14, 2026 announcing the seizure of 12 domain names, the office's allegations about AI-created non-consensual imagery of approximately 1,200 people, the DA's statement and the ongoing investigation. The release does not name the sites or any operator. Applies to da.

Read in English on 2026-10-07 (live HTTP 200; also read 2026-09-17 and 2026-10-06). WIRED, Sep 14, 2026 (Matt Burgess). Its 1,200 figure and the description of the victims are attributed to the DA's office and share that origin. Its own contributions are the reporter's observation of seizure notices before the announcement, archived-site observations and interviews with two researchers. The politician numbers it reports mix people mentioned with people who appeared and are not used as a count. Applies to wired.

Event countries: Unknown. Affected-person countries: Unknown. Court countries: United States.

Court US: the seizure was executed by the Manhattan District Attorney's Office under a court order, which WIRED reports as a New York State Supreme Court seizure warrant. Event unknown: no inspected source states where the depictions were generated or published or where the operators are, and the seat of the seizing authority does not locate the abuse. Affected people unknown: the depicted people are not enumerated, and WIRED's references to US, UK and European politicians mix people mentioned with people who appeared.

Reviewed for publication 2026-10-07: Published as a bounded series of alleged non-consensual AI-generated sexual depictions: the DA's release describes AI image and video tools turning real people's photos and videos into sexual imagery sold on 12 seized sites, and a researcher quoted by WIRED describes face-swapped pornographic videos on those sites. The 14 September 2026 announcement is documented; the allegations, the approximately 1,200 estimate and the researcher observations are attributed and the dates and count of the depictions remain unknown. No individual victim is named.

People described

People the Manhattan DA says were depicted without consent in AI-generated sexual imagery sold on the seized sites: approximately 1,200 by the office's estimate, overwhelmingly women and primarily public-facing figures such as actors, politicians, athletes, musicians, advocates and influencers. None is named in the inspected sources; ages and an exact count are not established.

People reported harmed in this case

Not reliably quantifiable from the sources

The DA's office gives an approximate estimate of 1,200 depicted people and WIRED attributes that figure to the office; neither source states an exact total, a lower bound or a counting method, so the numeric fields are zero placeholders for an unquantified count and the estimate stays in attributed prose. Twelve is the number of domains, not people. WIRED's politician numbers mix people mentioned with people who appeared and are not a count of depicted people. Audience figures (views) are not counted. The depicted people are not reported as users of the tools.

Counted once within this case. The same person may appear in other cases. This count does not establish AI causation.

Cite this case

Compiled per our published methodology: verification statuses, sourcing standards, and corrections process.

APA

NOPE. (2026). Manhattan DA says 12 seized websites sold AI-generated sexual deepfakes of about 1,200 people; the depictions' dates and an exact count are not established. AI incidents. https://nope.net/incidents/2026-manhattan-da-deepfake-ncii-site-seizure

BibTeX

@misc{2026_manhattan_da_deepfake_ncii_site_seizure,
  title = {Manhattan DA says 12 seized websites sold AI-generated sexual deepfakes of about 1,200 people; the depictions' dates and an exact count are not established},
  author = {NOPE},
  year = {2026},
  howpublished = {AI incidents},
  url = {https://nope.net/incidents/2026-manhattan-da-deepfake-ncii-site-seizure}
}

Related cases

High Unidentified image and video tool

Solo and Semarang, Indonesia: a university student's face was composited with AI onto pornographic images and a video by her ex-boyfriend and posted online; the suspect was detained after a nearly seven-month investigation

A female student at a state university in Solo (Surakarta), Central Java, learned on 1 January 2026 from friends that her face had been composited onto pornographic images and a video, which her lawyer said were found on Instagram and Telegram. Her family reported the case to the Central Java regional police cyber directorate on 28 January 2026. Police issued a formal police report on 31 July, arrested her ex-boyfriend, a university student in Semarang, at his boarding house in Gunungpati on 4 August 2026, and detained him. The police spokesman said the suspect used AI to place the complainant's face on another woman's naked body so that she appeared in pornographic content, uploaded it to his X account, did not sell it, and acted out of resentment after an on-and-off relationship ended. He faces charges under Articles 51 and 35 of the Electronic Information and Transactions Law and an article printed by two outlets as 'Law No. 1 of 2026', with a maximum of 12 years' imprisonment. Her lawyer said she could not sleep, withdrew, felt shame and at one point lost hope, that his team had found seven women in total allegedly targeted of whom six had not reported, and that the suspect's parents asked for an out-of-court settlement; police said only one complainant was confirmed.

High Unidentified image and video tool

Rohtak district, India: POCSO case registered after a 17-year-old says two men made obscene AI images of her and used them to coerce repeated rape since 2025

Police in Rohtak district, Haryana, registered a case under the Protection of Children from Sexual Offences Act and rape provisions, reported on 8 October 2026, after a 17-year-old girl from a village in the district told them that two men who delivered building material to her home had taken photographs of her, made obscene photographs and videos from them with AI, and used the threat of posting the material on social media to force her to meet them and rape her repeatedly, according to Dainik Bhaskar and Jagmarg News (both 8 October 2026). In the complainant's account relayed by the two outlets, the first assault followed an Instagram message in December 2025, further assaults took place in January and March 2026, and the most recent on 5 October 2026, after which she collapsed in a lane near her home and told her family, who called the police. The district Child Welfare Committee said she has been counselled. The police statement quoted by Dainik Bhaskar describes a complaint of forced sexual acts by two men and an investigation, without mentioning AI; the AI claim rests on the complainant's account as relayed by the outlets. No arrest is reported and the accused have not been heard.

High Unidentified image and video tool

Ahmedabad, Gujarat: a 27-year-old Delhi man was arrested for posting AI-morphed obscene images and videos of a young woman and her mother (more than 100 images found) on fake accounts in her name after she declined his friendship, while posing as the friend reporting the fake accounts for her

On 28 May 2026 the Ahmedabad Cyber Crime Branch announced the arrest in north-east Delhi of a 27-year-old man who posts religious discourse videos online. Police allege that after meeting a young Ahmedabad woman with a public social-media profile through religious conversations on Instagram and being refused friendship, he stalked her online from December 2025 to April 2026, created fake accounts in her name (three Instagram accounts and a YouTube channel at first; eight to ten across X, Facebook, Instagram and YouTube were found on his phone), downloaded photographs of her and her mother, used websites found by searching "AI remove clothes" to generate morphed obscene images and videos of the two women, and posted them with obscene captions to damage her reputation; investigators say more than 100 AI-morphed images of the complainant were found. Officers told ThePrint and ANI that she had confided in him about the accounts and that he told her to keep quiet and claimed to be reporting the accounts and getting hers reinstated. She complained to the Cyber Crime Police Station; an FIR was registered on 6 April 2026 under BNS Sections 78(2) and 356(2) and IT Act Sections 66(C) and 67, and the accused was traced through technical surveillance and arrested under the Branch's "Mission Cyber Rakshika" campaign. The complainant said she was mentally harassed and defamed; the DCP said the images also depicted her maternal aunt. Further investigation was under way; no charge sheet or court outcome was found.

Critical Unidentified image and video tool

Indian student dies after blackmail with sexual images police and family call AI-made

A 19-year-old student in Haryana, India, died in October 2025 after being blackmailed with nude images of himself and family members. His father says the images were made with AI from a hacked phone, and police called the case misuse of AI technology. The tool and the suspects remain unidentified in the reporting.

If you or someone you know is struggling, free and confidential support is available. Find a helpline near you at Signpost.