🏛️ Electoral authorities: adopt AI only where it's proportionate, and build the capacity to evaluate it independently.
🤖 AI chatbot providers: redirect electoral queries to official sources and ensure political balance.
What we found: parties and candidates are among the most prolific users of generative AI and much of that content goes unlabelled. In some cases, despite public commitments to label it.
The EU is developing guidance on AI use in elections. Here's what we told them ✍🏽
We submitted input to the European Commission's targeted stakeholder consultation under the European Democracy Shield, drawing on our research across 2024 European Parliament, 2025 German and 2026 Hungarian elections.
What we recommend 👇
🗳️ Political parties: label AI content, and stop using AI to manipulate or suppress voters.
👨🏽💻 VLOPs & VLOSEs: treat AI political content as high-risk, enforce labelling, and set clear sanctions for repeat offenders.
Voluntary commitments only work if they're kept. This guidance is a chance to make that the standard 💯
Voters, meanwhile, increasingly turn to AI chatbots for election guidance, only to get inaccurate or fabricated answers, with none reliably pointing them to official sources.
“Europe is not immune to democratic pressures, but we have active citizens,” says Paul Zoubkov, from @democracyreporting.bsky.social. “We have committed local governments, independent civil society and institutions that continue to invest in democratic renewal.”
Decentralised social media networks 🤔
What are they exactly?
How are they different from other platforms?
And could they be part of the answer to disinformation?
Hear from @aaron.bsky.team, @bsky.app's Head of Trust and Safety, who joined us at #DisinfoCon2025
