On 19 May 2025, the US President signed a new law named TAKE IT DOWN Act (S.146) (‘the Act’), which protects victims of real and deepfake ‘revenge pornography,’ a crime that is increasingly affecting women and teenage girls. This Act represents a significant legislative effort to address Technology-Facilitated Gender-Based Violence (‘TFGBV’), particularly concerning non-consensual intimate imagery and AI-generated deepfakes. Though it marks important progress, the Act also presents limitations that raise new human rights concerns, particularly regarding privacy and freedom of expression.

Deepfake-enabled TFGBV and human rights violation

TFGBV, as defined by the United Nations Population Fund, refers to gender-based violence (‘GBV’) enabled or amplified by digital technologies, including acts like deepfake pornography, which emerges as one of the most severe manifestations. Unlike traditional GBV, TFGBV leverages technology to perpetrate violence, which not only expands the scale and speed of harm but also deepens human rights violations.

First, human rights violations caused by traditional GBV may be exacerbated. Traditional harassment is often localised, but deepfake content can go viral globally within minutes due to the borderless nature of online platforms. It makes removal nearly impossible without strong legal protections. Second, deepfake-enabled TFGBV makes the legal enforcement to protect human rights more difficult. AI-generated abuse creates new forms of harm that are challenging to prove due to the difficulty of verifying the authenticity and origin of digital content, while the anonymity and cross-border nature of such crimes further intensify accountability problems. Third, the weaponisation of AI-generated scandals poses indirect threats to social and political rights, as fake content may be used to intimidate female politicians, activists, and journalists into silence, ultimately undermining democratic participation.

TAKE IT DOWN Act: legal responses and limitations

The Act recognises the publication of non-consensual deepfake pornography as a federal criminal offence and grants victims a legal pathway to seek damages and hold perpetrators accountable, filling a key legal gap. Additionally, it requires platforms to remove deepfake abuse within 48 hours of a valid request, ensuring swift action to reduce long-lasting reputational and psychological harm.

However, the Act also raises several human rights concerns. One critical concern is the tension between ensuring valid notifications and protecting the right to privacy. Section 3(a)(3) requires that a valid removal request be submitted by an identifiable individual or an authorised representative. However, victims of deepfake sexual abuse may fear public exposure, retaliation, or further harassment if their identities are disclosed during the take-down process. While the Act aims to protect the privacy rights of those affected by TFGBV, the current mechanism could inadvertently create new privacy vulnerabilities. To effectively safeguard victims, it is essential to address this conflict by considering alternative reporting mechanisms and providing supplementary safeguards – such as confidential or anonymous channels – that enable individuals to file take-down requests without incurring additional harm. This approach would ensure that the measures intended to protect victims do not, in turn, compromise their safety or privacy.

Another key concern is the potential conflict between the Act and protections for freedom of expression. The First Amendment guarantees individuals the right to freedom of speech, and any law imposing significant restrictions on speech must be narrowly tailored to avoid impeding constitutional rights. Article 19 of Universal Declaration of Human Rights and Article 19 of International Covenant on Civil and Political Rights also emphasise that any state-imposed content restrictions should be provided by law and respect the principles of necessity and proportionality. However, the Act lacks sufficient guardrails. For instance, it does not provide minimal speech protection. The broad definitions of digital forgeries and intimate visual depictions potentially include any images with intimate or sexual content without specifying exceptions for commercial use or public concern. Moreover, the Act neither incorporates any anti-abuse provisions nor establishes a mechanism for reviewing the legitimacy of take-down requests, potentially leading to an influx of frivolous or abusive requests.

Looking forward

The misuse of emerging technologies, particularly AI, raises significant human rights concerns, while existing legal frameworks may be inadequate and could inadvertently lead to further rights-related challenges. Looking forward, a more nuanced and balanced legal strategy that bridges AI governance and robust human rights protections is imperative to combat emerging forms of technology-facilitated violence effectively.