In short, technical capabilities ⬆️, preparedness ⬇️. And I haven't even mentioned how cuts to CISA make USG less prepared for AI-enabled interference efforts...

We can't get complacent just because last year's elections weren't wrecked by AI, but need to work hard to protect the midterms.

🪦 Info sharing: It helps support a info integrity and election security, yet it's on the outs. Though the 2024 AI Elections Accord wasn't perfect, it was a valuable starting point and framework for collaboration on provenance/detection and limiting misuse. The Accord's site has been removed:

🤖 Chatbot policies: Companies mitigated risks last year by temporarily limiting chatbot responses on certain political Qs. Now they're reducing guardrails on political outputs and disinfo. Given current political pressure/accusations of bias, what will their election policies look like in '26?

🎢 Unpredictable regulatory environment: 26 states have passed laws on AI in elections. Plus there's the federal level and AI moratorium efforts. As @travisrh.bsky.social said: Congress rejected the preemption, and the EO disregards states' democratic processes www.washingtonpost.com/technology/2...

Bad actors are also using AI more. Think AI-enabled influence ops, chatbots spitting out Kremlin-linked disinfo, or even taking advantage of elections for fraud schemes: www.cdmrn.ca/publications...

📈 Expect more AI: The biggest limitation on AI usage in campaigns last year was fear of backlash. But taboos are fading, and we see increased use of AI for political messaging (AKA to make a point, not to deceive).

More on campaign uses: cdt.org/insights/pro...

And important to remember that the premise is faulty - these changes to SAVE aren't needed for voter roll maintenance. Systems like ERIC already exist and are actually designed for that purpose

Political speech is indispensable and worth fighting for. At the same time, we all deserve an information ecosystem that’s transparent and empowers us to know who’s working (and paying) to win our votes and attention.

Voters deserve transparency and ethical campaigning. Influencers and political actors deserve clarity about best practices and what is allowed.

We offer recs to help govs, companies, and political actors support a trusted, rights-respecting info environment around elections cdt.org/insights/arc...

There's also a crossover with CIB, which is about to get worse as this industry collides with AI. As @emanuelmaiberg.bsky.social wrote last month: artificial influencers are coming www.404media.co/a16z-backed-...

And there's a tension: influencers support democratic participation/political discourse, but are also a vector for FIMI and operate in an opaque way that doesn't align w/ democratic principles like transparency.

We care about free speech - and we should also care whether the speech if freely given.

The TLDR: Platform policies are easily circumvented and/or difficult to enforce. Political influencers are too political to be governed by the FTC and too digital for the FEC.

Our report looks at the nitty gritty of relevant policy...

📱 Social media platform policies including branded content, political ads, and monetization

🏛️ US legal framework, namely FEC, FTC, FARA, state laws, 1A issues

...and finds that influencers fall through the cracks.

Political influencers’ importance has been growing for years, and that was really cemented in the 2024 election. Many many millions of dollars are going into this industry. (h/t to reporters like @makenakelly.bsky.social for bringing this issue to light) www.wired.com/story/influe...

Before I get to findings, huge shoutout to my coauthor @beccabranum.bsky.social + our CDT colleagues. Too many people touched this report to tag here, and I especially owe Becca for her patience with me not being a lawyer as we stitched together the many crosscutting issues in this report