1 in 6 Congresswomen Have Reportedly Been Targeted by AI-Generated Nonconsensual Intimate Imagery

A study by the American Sunlight Project is reported to have found that 1 in 6 Congresswomen were targeted by AI-generated nonconsensual intimate imagery (NCII) shared on deepfake websites. The study reports having found 35,000 mentions of explicit content involving 26 members of Congress, with 25 being women. Women were 70 times more likely than men to be victimized, according to the report.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.3 Fraud, scams, and targeted manipulation

The incident involves the targeted creation of humiliating and sexually explicit synthetic imagery of specific female lawmakers to harass and silence them.

Additional risk subdomains

• 1.2 Exposure to toxic content: The generation and hosting of deepfake pornography exposes the victims and internet users to highly toxic, nonconsensual sexual content.
• 1.1 Unfair discrimination and misrepresentation: The targeting disproportionately affected female lawmakers, representing a gender-based misrepresentation and harassment campaign.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The incident was driven by human actors intentionally using deployed generative AI tools to create and distribute nonconsensual explicit imagery targeting female lawmakers.

EU AI Act risk tier

• Risk tier: 3 Limited Risk

Limited Risk: The report describes 'AI-generated content such as deepfakes' which are subject to specific transparency obligations under the EU AI Act to ensure users are aware of the synthetic nature of the content.

AI system and alleged parties

• AI system: None named
• AI purpose: Deepfake Video Generation; Deepfake Image Generation
• Behaviour type: Tool
• Alleged developer: Unknown deepfake technology developers
• Alleged deployer: Unknown deepfake creators
• Alleged harmed parties: Congresswomen

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Substantial. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Substantial, indirect Negligible
• Differential treatment: direct Minor, indirect Substantial
• Civil rights: direct Minor, indirect Substantial
• Democracy: direct Minor, indirect Substantial
• Privacy: direct Minor, indirect Negligible
• Psychological: direct Minor, indirect Substantial
• Epistemic: direct Minor, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: Yes, the report describes over 35,000 mentions of lawmakers on deepfake pornography websites.

Directly caused: Over 35,000 instances of AI-generated nonconsensual intimate imagery depicting 26 lawmakers were hosted on 11 deepfake websites.

Inferred additional harm: It is highly likely that millions of similar toxic, nonconsensual sexual deepfakes depicting other individuals exist across these and other platforms.

Differential treatment

Reported: Yes, the report highlights a stark gender disparity where female lawmakers are disproportionately targeted compared to men.

Directly caused: Women members of Congress were 70 times more likely to be targeted than men, with 25 women and only 1 man depicted in the study.

Indirectly caused: The reports note a 'disparity of privilege' where lawmakers can get content removed quickly, while ordinary women lack the resources to do so.

Inferred additional harm: This gender-based targeting likely reinforces systemic barriers and discourages women from participating in public life and leadership roles.

Civil rights

Reported: Yes, the report describes the nonconsensual sexualization of women as an assault on women in leadership and their right to participate in public discourse.

Directly caused: The basic civil rights of 25 female lawmakers to engage in public service without gender-based sexual harassment were violated.

Indirectly caused: The fear of deepfake abuse discourages other women from running for office, undermining equal representation.

Inferred additional harm: Widespread proliferation of NCII threatens the fundamental rights of women to privacy, dignity, and equal participation in civic life.

Democracy

Reported: Yes, the report explicitly describes the targeting as a direct assault on democracy and free speech.

Directly caused: The harassment of 26 members of Congress undermines the integrity of democratic representation.

Indirectly caused: The threat of sexualized deepfakes causes 41 percent of young women to self-censor, silencing public speech and weakening democratic participation.

Inferred additional harm: The potential for blackmail and geopolitical concessions using deepfakes poses a broader threat to national security and democratic governance.

Privacy

Reported: Yes, the report describes the creation of nonconsensual intimate imagery as a violation of intimate privacy.

Directly caused: The intimate privacy of 26 lawmakers was violated by using their likenesses to generate explicit sexual content without consent.

Inferred additional harm: The lack of federal laws leaves millions of citizens vulnerable to severe violations of their intimate privacy through AI-generated content.

Psychological

Reported: Yes, the report describes devastating mental health effects, harassment, and shaming experienced by victims of image-based sexual abuse.

Directly caused: The 26 targeted lawmakers and ASP CEO Nina Jankowicz experienced harassment, distress, and the psychological toll of having their likenesses sexualized without consent.

Indirectly caused: The reports note that 41 percent of young women self-censor online to avoid harassment, indicating widespread indirect psychological distress and fear.

Inferred additional harm: It is highly likely that many other public figures and private citizens depicted on these sites suffer severe, unquantified anxiety, depression, and trauma.

Epistemic

Reported: Yes, the report describes the creation of believable synthetic media of events that never actually happened.

Directly caused: Over 35,000 synthetic, false depictions of lawmakers in sexually explicit scenarios were generated and distributed.

Inferred additional harm: The widespread availability of highly realistic deepfakes erodes public trust in visual media and complicates the verification of real events.

People affected

• Occurrences reported: 1
• People reportedly harmed: 26
• People reportedly exposed: 26

Potential causes

Management

• Failure of Tech Self-Regulation: Critics lack optimism in Big Tech's ability to self-regulate platform harms.
• Inadequate Platform Action: Platforms fail to proactively prevent the hosting and sharing of NCII.

Technology

• Generative AI Accessibility: Generative AI makes it easy for perpetrators to create realistic NCII.
• Persistent Search Indexing: Google search pages remain indexed even after deepfake content is removed.
• Machine Learning Realism: Advanced ML creates highly believable fake pictures, audio, and text.

Data Inputs

• Public Availability of Headshots: Headshots of public figures are easily sourced to overlay on adult media.
• Target Name Queries: Custom search engines easily locate targets using names and nicknames.

Human Factors

• Gender-Based Harassment Motive: Perpetrators seek to shame, strip, and silence women in public life.
• Self-Censorship of Victims: Fear of online harassment causes women to self-censor and avoid public roles.

Process and Methods

• Ineffective Platform Moderation: Platforms like Instagram fail to act on reported abusive comments.
• Unequal Takedown Capabilities: Private citizens lack resources to achieve rapid content removal.
• Lack of Control Over Proliferation: Even after takedowns, content continues to proliferate without control.

Regulatory Environment

• Absence of Federal Laws: No federal law establishes criminal or civil penalties for AI-generated NCII.
• Patchwork of State Laws: State laws are inconsistent, mostly offering civil instead of criminal paths.
• Legislative Gridlock: Bipartisan bills like DEFIANCE and TAKE IT DOWN face hurdles in the House.

Information quality

• Classification confidence: High
• Reason for confidence: The reports are highly consistent, citing the same American Sunlight Project study with identical statistics (35,000+ mentions, 26 lawmakers, 25 women, 1 man). The details regarding the social and political impacts, legislative efforts, and the gendered nature of the harm are clearly documented across multiple reputable sources.
• Ambiguities identified: The exact generative AI models or platforms used to create the deepfakes are not identified, and the precise mechanism behind the rapid removal of the content remains unexplained.
• Alternative interpretations: The incident could be viewed purely as a targeted harassment campaign rather than an AI safety failure, but the central role of generative AI in enabling the scale and realism of the abuse makes the AI safety classification primary.

An investigation revealed over 35,000 AI-generated deepfakes targeting 26 members of the US Congress, disproportionately affecting female lawmakers. While not a direct threat to critical infrastructure or state sovereignty, the incident represents a substantial threat to societal stability and human rights, highlighting how generative AI can be weaponized for targeted political harassment and gender-based discrimination.

• Overall national security impact: Substantial
• Scope: Single nation
• Primary target: United States
• Alleged perpetrator: Unknown

Threat characteristics

• Imminence: Long-term. The incident represents an ongoing strategic concern regarding the proliferation and regulation of generative AI tools rather than an active tactical crisis.
• Autonomy: Human-controlled. The AI models were utilized strictly as tools by human perpetrators to generate and distribute synthetic media.
• Novelty: Evolved capability. While deepfakes are an established threat, the scale, automation, and systemic targeting of national political leadership represent a significant evolution in harassment tactics.

Impact by dimension

• Physical security: Negligible. No physical threat, kinetic targeting, or critical infrastructure manipulation occurred in this incident.
• Information security: Minor. Involves deepfake operations targeting national political leadership, but currently characterized as targeted harassment rather than a coordinated state-sponsored intelligence or disinformation campaign.
• Sovereignty: Minor. Targeting of 26 members of Congress with nonconsensual imagery creates pressure on public officials but has not disrupted core constitutional processes or state authority.
• Economic security: Negligible. No significant impact on financial systems, strategic technology theft, or economic stability was reported.
• Societal stability: Substantial. The targeted generation of over 35,000 deepfake pornographic images disproportionately targeting female lawmakers represents a severe violation of intimate privacy and systemic gender-based harassment that threatens equal participation in public life.