AI-assisted online harassment including deepfake imagery and manipulated content has been steadily on the rise and women are targeted at rapidly increasing rates. AI generated or augmented attacks are often used to undermine women’s professional credibility and cause reputational damage.

Italian Prime Minister Giorgia Meloni was subjected to one such harassment campaign which saw risque AI generated images of the politician circulated through social and legacy media. Meloni fought back against the images but noted that other women may not have the same resources to do so.

Her concern was reflected by a recent UN Women survey across 119 countries which indicated that nearly one in four women in public facing roles reported AI-assisted online violence. These concerning statistics could potentially be improved due to an article in the EU AI Act that went into effect this past summer. Article 50 established obligations for providers to mark synthetic audio, image, video and text outputs in machine-readable, detectable formats.

The emerging authentication technology entering the market takes a few distinct approaches to solving the problem of undisclosed, manipulated content. From the hardware side, producers like Apple, Nikon, and Canon have included on-device authentication to be able to indicate if an image was captured rather than generated.

Software giants like OpenAI, Google, and Anthropic have embedded machine-readable watermarks into their outputs to facilitate detection of generated content.  Institutions like C2PA, The Coalition for Content Provenance and Authenticity, have established open standards for a verifiable chain of custody for digital content in order to track file manipulation.

And if manipulated files do find their way onto the internet, hosting platforms like YouTube have established programs to detect AI impersonations of public figures and undisclosed AI generated content. The EU AI Act’s transparency requirements may have accelerated the adoption of existing developments of authentication technology but a valuable consequence is the potential protection of women from the concerning trend of AI-assisted harassment.