EU AI Act Revisions Aim to Ban Deepfakes to Protect Women

AlgorithmWatch has recommended the implementation of a deepfake ban within the ongoing revisions of the EU AI Act, targeting the rising misuse of AI for digital sexualized violence. This proposal responds to the coordinated inadequacies within the EU's regulatory framework, which currently lacks cohesion. Historically, efforts such as the Digital Services Act (DSA) have aimed to regulate content dissemination, but direct action against specific AI abuses, like non-consensual deepfakes, marks a more targeted regulatory evolution.

The proposed deepfake ban stands to shift power dynamics by holding AI companies directly accountable for generative misuse, a shift from current broadly dispersed liability. This could disadvantage platforms that have prospered under looser regulations while protecting victims who suffer from psychological and societal impacts. Furthermore, AI tool developers may need to enhance compliance mechanisms, impacting design and distribution strategies.

The EU legislative bodies are likely to integrate these recommendations into the revised AI Act by early 2027. Companies operating within Europe will need to adapt their models to comply with user safeguarding stipulations or face significant penalties. This policy shift hints at a broader EU trend towards specific and enforceable digital rights protections, potentially influencing other regions to follow suit.

This ban could lead to increased demand for compliance technologies and AI auditing firms, catalyzing a niche regulatory tech market. Simultaneously, it might strain global platforms in harmonizing different regional regulations, leading to potential service withdrawal or tailored country-specific features.