The European Commission is increasing pressure on major online platforms to explain how they are preparing for risks linked to generative AI. The move comes ahead of the June European Parliament elections and targets some of the largest services used to create, share and amplify online content.
The formal requests for information were sent to Google, Meta, Microsoft, Snap, TikTok and X. They concern Bing, Facebook, Google Search, Instagram, Snapchat, TikTok, YouTube and X, all of which are designated as very large online platforms under the Digital Services Act.
Why the EU is asking now
The Commission is using the Digital Services Act to ask these platforms for details about their mitigation measures for generative AI risks. Under the DSA, very large online platforms must assess and reduce systemic risks, while also complying with the wider rulebook.
The requests focus on both the creation and the spread of generative AI content. That distinction matters because a platform can be involved in more than one part of the risk: it may host AI-generated material, recommend it, help it travel quickly, or provide tools that make it easier to produce.
The Commission identified several areas of concern. These include so-called hallucinations, where AI technologies generate false information; the viral spread of deepfakes; and automated manipulation of services in ways that can mislead voters.
The election context gives the requests urgency. The Commission noted that the cost of producing synthetic content is going down dramatically, increasing the risk that misleading deepfakes could be produced and circulated during elections.
What platforms must explain
The RFIs ask the companies to provide information about how they are dealing with generative AI risks on their services. The Commission is also asking for internal documents linked to risk assessments and mitigation measures.
The areas covered go beyond election security. According to the Commission, the requests address risks related to:
- the impact of generative AI on electoral processes
- the dissemination of illegal content
- the protection of fundamental rights
- gender-based violence
- the protection of minors
- mental well-being
That broader scope shows that the Commission is not treating generative AI as only a political advertising or campaign issue. It is also looking at how synthetic media can affect safety, rights and harmful content across large online services.
Some of the risks are tied to images, video or audio. The Commission’s questions are intended to cover malicious synthetic content, including deepfake porn and other forms of harmful AI-generated material.
Election security is the urgent part
The platforms have been given until April 3 to provide information related to the protection of elections. The Commission described that part of the request as urgent.
At the same time, the EU hopes to finalize election security guidelines for very large online platforms by March 27. The current requests are partly intended to support that guidance.
The Commission is also planning stress tests after Easter. Those tests are expected to examine how ready platforms are to deal with generative AI risks, including the possibility of a flood of political deepfakes before the June European Parliament elections.
A senior Commission official, speaking on condition of anonymity, said the EU wants platforms to explain what they are doing to be as prepared as possible for incidents that may need a response in the run-up to the elections.
Election security has already been named as one of the priority areas for enforcement under the DSA rules that apply to very large online platforms. The Commission has also recently been consulting on election security rules as it works toward formal guidance.
How this fits into the EU’s wider approach
The EU does not view voluntary industry action as enough on its own. A tech industry accord to combat deceptive use of AI during elections came out of the Munich Security Conference last month, with support from several of the same platforms now receiving RFIs. In the Commission’s view, that accord does not go far enough.
A Commission official said the forthcoming election security guidance will go much further. The EU plans to draw on several layers of safeguards: the DSA’s due diligence rules, more than five years of work with platforms through the Code of Practice Against Disinformation, and transparency labelling and AI model marking rules expected under the incoming AI Act.
The Commission’s stated aim is to build an ecosystem of enforcement structures that can be used in the run-up to elections. In practice, that means combining formal oversight of very large online platforms with existing self-regulatory work and the future framework expected under the AI Act.
The RFIs also reflect other DSA enforcement priorities, including illegal content such as hate speech and child protection. For the broader generative AI questions not limited to election protection, platforms have until April 24 to respond.
Why smaller services still matter
The Commission’s direct DSA oversight in this context applies to designated very large online platforms. Smaller platforms and smaller AI tool makers are not covered in the same way if they are not designated.
Even so, the EU is watching them as part of the same risk landscape. Misleading, malicious or otherwise harmful deepfakes may be distributed on smaller services, while smaller AI tools can make it easier to generate synthetic media at lower cost.
The EU’s strategy is to widen the regulatory impact indirectly. Larger platforms can act as amplifiers or distribution channels, so pressure on those services may affect how synthetic content travels across the broader internet.
The Commission is also looking to self-regulatory mechanisms such as the Code of Practice Against Disinformation and the AI Pact. The AI Pact is expected to get up and running shortly once the AI Act is adopted, which is expected within months.
The immediate message to the largest platforms is clear: the EU wants detailed answers before the election period intensifies. Generative AI has made synthetic content cheaper and easier to produce, and the Commission is now using the DSA to test how prepared major platforms are to manage the risks.