Key claims
The deepfake was extremely convincing. The synthetic video showed someone appearing nearly identical to Hunter engaging in pornographic activity. Despite Hunter knowing the video was impossible, the visual realism created visceral fear: family members questioned whether it was authentic, and strangers approached her on the street based on the video's false premise.
Nonconsensual intimate imagery (NCII) deepfakes target women disproportionately. Hunter notes that six months before the campaign video, she received 15 AI-generated images of herself in lingerie. Her conversations with other female politicians revealed a pattern: this type of abuse is routine for women in public life, creating a chilling effect on women's political participation.
Timing and scale amplified political harm. The video circulated during the final three weeks before the election, when Hunter had limited time to counter the narrative. Messages flooded in from across the country, across the Atlantic, and internationally, with strangers using the fake video as justification for harassment and sexual proposals.
The lie spread too far and too fast to effectively counter. By the time Hunter understood the video was AI-generated and tried to correct the record, rumors had penetrated her own family, community, and even reached elderly voters with limited media literacy. Traditional channels (police, platforms) were unhelpful or unreachable.
This is a case where misogyny and technology intersect. Hunter emphasizes that deepfake attacks on women are not gender-neutral technical harms—they combine sexual objectification, humiliation, and the power imbalance between the victim and her attacker. Deepfakes of male politicians (Nancy Pelosi, Joe Biden, President Zelensky) exist, but they are political in nature; deepfakes of female politicians are overwhelmingly sexualized.
Truth is foundational to democracy. Without trust in truth, democracy becomes vulnerable to manipulation and corruption. Deepfakes erode the epistemic commons and make it harder for voters to make informed decisions. When synetic media can undermine years of trust-building in weeks, the democratic process itself is threatened.
Deepfakes in politics are not yet illegal in most jurisdictions. Hunter was unable to pursue legal recourse against the perpetrator. This lack of legal consequence creates an asymmetry: the attacker faces no risk, the victim faces real harm.
Media authentication and detection are incomplete solutions. Hunter did not know whether the video was AI-generated or a sophisticated composite at first viewing. Even experts struggle to verify content in real-time. Detection requires technical sophistication that ordinary voters do not possess.
Key concerns raised
Gender-based political violence through synthetic media. The talk catalogs a pattern: female politicians receiving targeted NCII deepfakes while their male counterparts receive political-content deepfakes. This asymmetry is driven by misogyny and suggests that deepfakes are being used as a tool of gender-based harassment to suppress women's political participation.
Chilling effect on women's political participation. Hunter notes that women considering entry into politics see the fate of women already in office and rationally conclude that the harassment is not worth it. This represents a form of democratic harm: talented women are discouraged from public service by fear of synthetic-media abuse.
Vulnerable populations are disproportionately at risk. Hunter notes concerns about elderly voters with limited media literacy who may be more susceptible to believing fake videos. Younger cohorts saturated with AI-generated images may conversely become unable to distinguish real from synthetic media.
Law enforcement is under-resourced. When Hunter contacted local police, they informed her they lacked cybercrime capabilities to investigate. This suggests that reporting infrastructure is inadequate to address the scale of NCII deepfake attacks.
Solutions Hunter advocates
Ethical embedding in AI development. Rather than accepting deepfakes as inevitable, Hunter argues that AI researchers and companies must embed ethics, eliminate bias, install empathy, and ensure alignment with human values during design and deployment.
Regulation and policy intervention. Policymakers must pass laws making nonconsensual intimate imagery (including AI-generated versions) illegal, with real criminal penalties.
Public education and media literacy. Voters must understand the prevalence and sophistication of deepfakes to become more skeptical of extraordinary claims and encourage verification before sharing.
Platform accountability. Social media companies must enforce policies against the distribution of deepfakes and NCII, and must prioritize content moderation over engagement metrics.
Tech for good. Hunter expresses hope that technologists, regulators, and policymakers can work together to ensure that AI serves humanity rather than undermines democratic trust.
Notes
This talk is valuable as a first-person account of deepfake-enabled political abuse targeting a woman in a real election context. Unlike academic or expert-framed talks, Hunter centers the victim's experience: the psychological impact, the family shame, the social alienation, the paralysis of not knowing who created the video or why.
The talk is also notable for naming the intersection of misogyny and technology as its own category of harm. The deepfakes of male politicians (Pelosi, Biden, Zelensky) discussed in the talk are framed as political manipulation; the deepfakes of Hunter are framed as sexual humiliation. This distinction is crucial for understanding that deepfake abuse is not a gender-neutral technical problem.
Hunter's emphasis on the chilling effect is particularly important for democratic theory: synthetic media can suppress political participation by making the cost of candidacy (harassment, humiliation, psychological harm) too high for some groups. This represents a form of democratic exclusion that goes beyond individual harm.
The talk also illustrates the gap between technical detection and real-time verification. While digital forensics can expose deepfakes retrospectively, voters in the moment do not have forensic tools, do not have time to investigate, and do not have the expertise to verify. This asymmetry suggests that detection alone cannot be the solution; instead, upstream prevention (legal penalties, platform enforcement, AI safety practices) is necessary.
Finally, the talk demonstrates the inadequacy of current institutional responses. Police lack resources; platforms lack enforcement; legislatures lack laws; tech companies lack incentives. Hunter's call for "tech for good" acknowledges that the problem requires action at multiple levels simultaneously.