EU launches formal Digital Services Act probe into X over deepfakes

The European Commission announced a formal investigation under the Digital Services Act into X, the platform formerly known as Twitter, after researchers and EU regulators found that Grok, the chatbot developed by xAI and integrated into X, had been used to generate a large volume of sexually explicit and non-consensual images. The move, announced on January 26, 2026, raises fresh questions about how social networks deploy generative AI and how platforms police emerging harms.

The probe is one of the Commission’s most significant uses of the DSA’s enforcement powers to date. The law imposes obligations on very large online platforms to assess and mitigate systemic risks, prevent dissemination of illegal content, and ensure transparent content moderation. Investigators will examine whether X complied with those obligations in design, operation, and enforcement of safeguards around Grok and user-facing systems that allow image creation and sharing.

According to the Commission’s statement, evidence from researchers and regulatory bodies indicated that Grok had been used to produce a substantial volume of sexualized deepfakes that depicted non-consenting individuals. The inquiry will focus on whether X’s systems amplified, failed to moderate, or otherwise allowed the spread of this harmful content inside the European Union, and whether the platform’s risk assessments and mitigation measures were adequate under the DSA.

Under the Digital Services Act, serious infringements of platform duties can lead to corrective orders, mandatory changes to product design and moderation practices, and financial penalties. For the most severe breaches, fines can reach substantial portions of a company’s global turnover. The Commission can also require ongoing compliance audits and impose structural measures intended to reduce future risks to users.

Beyond potential enforcement, the investigation will test how regulators evaluate AI components embedded in social networks. Grok’s presence inside X blurs the line between conversational tools and content-generation services, complicating responsibilities for content that is created by or with the assistance of an on-platform AI. Regulators are expected to probe technical safeguards such as rate limits, red-team testing, content filters, provenance and watermarking of images, and transparency about model capabilities and limitations.

The case highlights a broader tension facing platforms and suppliers of generative models: how to balance innovation and user convenience against the potential for rapid misuse that can harm individuals and communities. Non-consensual sexual imagery and deepfakes inflict real-world damage, privacy violations, harassment, reputational harm, and often fall into legal gray areas that existing moderation systems struggle to address quickly at scale.

The Commission’s formal procedure will give regulators access to X’s internal documentation and decision-making processes related to Grok and associated moderation. Depending on findings, the inquiry could result in orders that change how AI features operate in the EU or stricter oversight of how models are integrated into social networks.

As this investigation proceeds, it will likely influence industry practices for integrating generative AI into mass-market platforms and shape legislative approaches to AI safety and platform responsibility across Europe and beyond. The outcome may set new precedents for how regulators hold technology companies accountable for harms that emerge from advanced, user-facing AI.