AI Impersonation Prevention Act of 2025
- Sponsor
- Rep. Ansari, Yassamin [D-AZ-3]
- Committees
- Judiciary Committee (primary)
- Last action
- Jul 23, 2025
Bottom line
The bill criminalizes the malicious use of AI to impersonate federal officials, aiming to prevent disinformation while safeguarding legitimate expressive content through clear disclosure requirements.
What it actually does
This bill amends Section 912 of Title 18, United States Code, to specifically prohibit the knowing use of artificial intelligence to impersonate a federal officer or employee, including by mimicking their voice or likeness, when such action produces materially false or misleading content. It includes an explicit carve-out for legitimate uses in satire, parody, or expressive conduct protected under the First Amendment, provided such content includes a clear disclosure of its inauthenticity. The bill also defines 'artificial intelligence' and 'impersonates' for the purposes of this prohibition.
Proponents argue
Supporters argue this bill is a critical and timely measure to protect the integrity of government communications and prevent the spread of AI-generated disinformation that could undermine public trust in federal officials. They contend it updates existing impersonation laws to address the sophisticated threats posed by deepfakes and other generative AI technologies, ensuring accountability for malicious actors and safeguarding the public from deceptive content.
Opponents contend
Critics might express concerns that the bill's definitions, particularly 'materially false or misleading content' or 'reasonably likely to cause another person to believe the content is authentic,' could be overly broad or subjective. They might argue this could potentially lead to unintended consequences, such as chilling legitimate speech or leading to inconsistent enforcement, despite the explicit carve-out for satire and parody.
The bill is exceptionally concise and clear, allowing for a thorough evaluation within a short timeframe.
Section 2, amending Section 912 of title 18, United States Code
Prohibition on AI-Based Impersonation of Federal Officials
This provision adds a new subsection (b) to the existing federal law against impersonating federal officials. It makes it a crime to knowingly use artificial intelligence to falsely assume or pretend to be a U.S. officer or employee, specifically by mimicking their voice or likeness, if this produces content that is materially false or misleading. The provision explicitly states that legitimate satire, parody, or other First Amendment-protected expressive conduct is not prohibited, provided it includes a clear disclosure that the content is not authentic.
Supporters argue
Supporters argue this provision is essential for modernizing federal law to combat the growing threat of deepfakes and AI-generated disinformation. They contend it provides law enforcement with necessary tools to prosecute those who maliciously use AI to deceive the public and undermine government authority, while carefully carving out protections for free speech.
Critics contend
Critics might argue that the language defining 'materially false or misleading content' or 'reasonably likely to cause another person to believe the content is authentic' could be interpreted broadly, potentially leading to a chilling effect on legitimate speech or creating ambiguity in enforcement. They may also question the practical effectiveness of disclaimers in preventing deception in a rapidly evolving digital landscape.
Tradeoffs
This provision navigates the tension between the urgent need to combat AI-driven disinformation and the fundamental protection of free speech. The challenge lies in precisely defining the line where expressive conduct becomes criminal impersonation, especially as AI technology continues to advance and blur the lines between reality and fabrication.
This subsection defines 'artificial intelligence' as any system or software performing tasks normally requiring human intelligence, including generative models capable of producing human-like audio, video, or text. It also defines 'impersonates' as falsely representing oneself as another identifiable individual in a manner reasonably likely to cause another person to believe the content is authentic.
Section 2(c), amending Section 912(c) of title 18, United States Code
Why it matters:This is standard legislative practice to ensure clarity and precision in legal terms, rather than an attempt to obscure information.
Case for: Proponents argue these definitions are vital for providing clarity and scope to the new prohibition, ensuring it applies effectively to current and future AI technologies while specifying the necessary intent for an act to be considered impersonation.
Case against: Opponents might contend that the broad definition of 'artificial intelligence' could inadvertently encompass technologies not intended for malicious impersonation, or that the subjective nature of 'reasonably likely to cause another person to believe the content is authentic' could lead to inconsistent legal interpretations.
Estimated impact: These definitions directly determine the scope and enforceability of the AI impersonation prohibition, influencing how law enforcement, courts, and AI developers understand and comply with the law.