
US campaigns now run on AI at nearly every step, and Europe is drawing a harder line
Quick Answer
AI has become integral to US political campaigns, with 87% of strategists using it daily for tasks like data analysis and targeted messaging.
Quick Take
AI has become integral to US political campaigns, with 87% of strategists using it daily for tasks like data analysis and targeted messaging. In contrast, Europe enforces stricter regulations on AI in political advertising, requiring transparency and consent for data use, effective from October 2025.
Key Points
- 87% of US campaign strategists use AI daily for various tasks.
- Democrats show more skepticism towards AI compared to Republicans.
- EU regulations mandate clear labeling of political ads and consent for data use.
- Germany's CDU used an AI bot named 'Conrad' for drafting press releases.
- Self-regulation in Germany includes a fairness agreement among major parties.
📖 Reader Mode
~4 min readThe NYT report paints a picture of a deep shift in American campaigning. According to a survey by the newsletter Anchor Change, 87 percent of campaign strategists now use AI daily. Beyond the publicly visible layer of AI-generated images and videos, campaign managers have woven the technology into nearly every workflow, the New York Times reports. They're analyzing voter data, producing campaign materials, and crafting tailored messages for micro-segments of the electorate. The Democratic opposition research group American Bridge 21st Century used AI to vet roughly 250 Republican candidates, according to the report.
Both parties are all in on AI, but they play by different rules
Despite the boom, AI remains a political risk, the New York Times reports. Polls show that Democratic voters are more skeptical of the technology than Republicans. Progressive groups report angry emails about AI use, and unionized staffers worry about their jobs. Republican strategists, by contrast, face less internal pushback.
"If voters don't like A.I., they don't want to know that their candidate's campaign is using A.I. to do stuff like draft emails or create press releases or edit videos. So you're just not going to see people bragging about it. But it is happening," Republican strategist Eric Wilson, director of the Center for Campaign Innovation, told the New York Times. Wilson considers AI-generated videos of opponents acceptable as long as they reflect real statements. The National Democratic Training Committee, on the other hand, rejects such content outright because it "undermines democratic discourse and voter trust," according to the report.
Republicans lean more heavily on privately funded companies, while Democrats prefer nonprofit models, the NYT reports. Democratic hesitation could slow adoption and give Republicans an edge in tight races. The November midterms are widely seen as a test run for the AI strategies that will shape the 2028 presidential campaign.
Europe's rules for AI in campaigns are stricter
While US campaigns fold AI into their operations with little regulation, the EU is betting on transparency requirements and data protection. Since October 2025, new rules for political advertising apply across the EU. Political ads must be clearly labeled and disclose, among other things, who paid for them, which election they target, and how much money was spent. Political targeting requires explicit, separate consent from the people involved. Sensitive data like political views or ethnic background can't be used for profiling.
The AI Act will also matter for future campaigns. Transparency requirements for generative AI take effect on August 2, 2026. After that date, deepfakes and AI-generated or AI-manipulated texts on topics of public interest must be clearly labeled in key cases. The European Commission published a voluntary code of practice for labeling AI-generated content in June 2026.
AI had already arrived in Germany's 2025 federal election campaign, though on a smaller scale than in the US. According to ZDFheute, parties used AI for tasks like writing text, editing images, analyzing data, and managing social media. The CDU in North Rhine-Westphalia, for example, deployed an AI bot called "Conrad" to help staffers draft press releases and social media posts. The SPD, Greens, and Left Party named AI as a writing aid, while the FDP and Left Party used it for image editing. The AfD used it for graphics and select video clips.
In Germany, self-regulation also played a more explicit role. In December 2024, the CDU, CSU, SPD, Greens, FDP, and Left Party signed a fairness agreement. They pledged to clearly label AI-generated images, video, and audio, and committed to not using deepfake technology to put words in opponents' mouths.
The AfD and BSW did not sign on. The AfD was classified by Germany’s domestic intelligence agency in May 2025 as a “confirmed right-wing extremist” organization, though the classification has been temporarily suspended pending a final court ruling. BSW, a left-wing populist party, combines left-leaning economic policies with a restrictive stance on immigration and a foreign policy that opposes arms deliveries to Ukraine, calls for negotiations with Russia, and is sharply critical of NATO.
The AfD, which Germany’s domestic intelligence agency classified as a “confirmed right-wing extremist” organization in May 2025, a label that has been temporarily suspended pending a final court ruling, and BSW, a left-wing populist party known for pairing left-leaning economic policies with hardline anti-immigration positions and a pro-Russia, NATO-skeptical foreign policy, did not sign on.
— Originally published at the-decoder.com
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from The Decoder
See more →
An AI model programmed nonstop for 19 days on a single MirrorCode task that cost $2,600 to run
Epoch AI's MirrorCode benchmark reveals Claude Opus 4.7 as the leader with a 56% solve rate, reconstructing a 16,000-line toolkit in 14 hours. Despite this, all models tested struggle with the most complex tasks, highlighting limitations in current AI capabilities. The single task consumed $2,600 over 19 days, raising questions about cost-effectiveness in AI development.

