Statutes
Which of these laws has ever been used?
Legislatures have passed many laws aimed at AI-enabled conduct. This page shows which of them have led to a criminal conviction and which older laws prosecutors have used instead.
So far, laws written for AI have produced one located conviction, under the US TAKE IT DOWN Act. The other convictions in this record came under general criminal law, some of it older than the internet. South Korea's deepfake provisions are the exception: they are being used at volume.
A count of zero does not mean a law is useless. Most of these laws are recent, and prosecutors tend to use the charges they know. The count describes what the public record shows today.
TAKE IT DOWN Act
Non-consensual intimate imagery and digital forgeries of identifiable people.
One conviction located. A guilty plea in April 2026 led to a 180-month sentence in September 2026. The forgery count reached the adult victims; the child-related conduct in the same case was charged under a statute from 2003.
AI Basic Act
Governance, transparency and risk obligations on developers and deployers.
Regulatory rather than criminal. Promulgated in January 2025 and in force since January 22, 2026. Its only criminal penalty covers leaks of confidential information by committee members; other duties carry administrative fines of up to 30 million won. No enforcement action located. Listed because it is often mentioned in discussions of AI crime, although it creates no offence for misusing AI.
State AI election deepfake laws
Synthetic media depicting candidates, typically within a window before an election.
No criminal conviction located in any state. The most prominent prosecution, of the consultant behind a 2024 robocall imitating President Biden, ended in acquittal in June 2025 (NHPR); a separate $6 million FCC fine was regulatory, not criminal (FCC). Federal courts struck down California's law in August 2025 (Bloomberg Law) and Hawaii's in January 2026 (Reason). In September 2026 a federal judge found Montana's law likely unconstitutional in a ruling limited to the plaintiffs; the law remains in force (Daily Montanan).
Act on Special Cases Concerning the Punishment of Sexual Crimes
Production and distribution of sexually explicit synthetic media.
In a national crackdown from November 2024 to October 2025, fabricated sexual imagery was the largest category of cybersexual crime: 1,553 of 4,413 recorded cases. Counting deepfake-made child sexual exploitation material as well, police acted on 1,462 deepfake cases and arrested 1,438 people, 62% of them teenagers. These are arrest figures; conviction totals were not located. No other jurisdiction in this list is enforcing at comparable volume (Newsis, reporting Korean National Police Agency figures, November 16, 2025).
EU AI Act
Prohibited practices and risk tiers for AI systems.
Regulatory, phased, and enforced against providers and deployers rather than offenders. Criminal enforcement remains a member-state matter. No criminal prosecution located under national implementing provisions.
18 U.S.C. §§ 371 and 1343 (conspiracy to commit wire fraud)
Schemes to defraud using interstate wire communications.
The wire fraud statute dates from 1952. It carried the largest AI-enabled fraud conviction located to date: a guilty plea in March 2026 to conspiracy to commit wire fraud (charged under § 371, maximum five years) over hundreds of thousands of AI-generated songs streamed by bot accounts. The defendant was sentenced in October 2026 to 18 months and forfeited $8,091,843.64. Prosecutors used no AI-specific law.
18 U.S.C. § 1466A
Obscene visual representations of the sexual abuse of children.
Two decades older than generative AI and expressly carved out of the TAKE IT DOWN Act's intimate-depiction offence. It, not the newer statute, reached the child-related conduct in the first TAKE IT DOWN Act case.
C.R.S. § 18-6-403 (Colorado)
Sexual exploitation of a child, including production and possession.
Twenty-three counts brought in January 2026 over images said to have been produced by altering photographs of real children with generative tools. There are no federal counts. No disposition located.
720 ILCS 5/11-20.4 (Illinois)
Obscene depictions of a purported child, including computer-generated images and video.
Thirteen counts approved in October 2026 against an Antioch man, with more charges expected. No disposition located.
Fla. Stat. § 827.072 (Florida)
Generated sexual images of a fictitious person who appears to be a minor.
Two Martin County cases located, from February 2025 and September 2026. No disposition located in either.
What the counts mean
A number is the count of criminal convictions this record has located under that statute for AI-enabled conduct. In use marks a general statute that has carried such a conviction without being written for it. n/a marks a regulatory instrument with no criminal provisions to count. High volume marks a statute being applied at a scale where individual cases are not separately tracked here.
Absence of a located conviction is not proof that none exists. Court records are uneven across jurisdictions and most are not searchable in English. Where a conviction is found, this page will change and the entry it came from will be linked. Corrections to jason@unio-analytics.com.
Charged marks a statute with a case pending and no conviction yet.