Anthropic offers Claude AI election guarantees ahead of the US midterm elections

Anthropic, the AI ​​company behind the chatbot Cloud, Announced Friday A set of new election integrity measures designed to prevent its AI from being used as a weapon to spread misinformation or manipulate voters ahead of the 2026 US midterm elections and other major contests around the world this year.

The San Francisco-based company has detailed a multi-pronged approach that includes automated detection systems, pressure testing against influence operations, and a partnership with a nonpartisan voter resource organization — measures that reflect increasing pressure on AI developers to monitor how their tools are used during election seasons.

Anthropic’s usage policies prohibit using Claude to run deceptive political campaigns, create fake digital content intended to influence political discourse, commit voter fraud, interfere with voting infrastructure, or spread misleading information about voting processes.

To implement these rules, the company said that it subjected its latest models to a set of tests. Using 600 prompts – 300 malicious requests combined with 300 legitimate requests – Anthropic measured how reliably Claude complied with appropriate requests and rejected problematic requests. Cloud Opus 4.7 and Cloud Sonet 4.6 responded appropriately 100% and 99.8% of the time, respectively.

The company also tested its models against more sophisticated manipulation methods. Using multi-turn simulated conversations designed to reflect step-by-step methods bad actors might use, Sonnet 4.6 and Opus 4.7 responded appropriately 90% and 94% of the time when tested against influence operations scenarios.

Anthropic also tested whether its models were able to execute influence operations autonomously, that is, plan and execute a multi-step campaign from start to finish without human prompting. The company said that with warranties in place, its latest models declined almost all tasks.

On the issue of political neutrality, the company conducts assessments before launching each model to measure the consistency and fairness of Claude’s engagement with claims that express viewpoints from across the political spectrum. The Opus 4.7 and Sonnet 4.6 scored 95% and 96%, respectively.

For users looking for voting information, Claude will display an election banner directing them to TurboVote, a nonpartisan resource from Democracy Works that provides reliable, real-time information about voter registration, polling locations, election dates, and ballot details. A similar banner is scheduled to be raised for the Brazilian elections later this year.

Anthropic said it plans to continue monitoring its systems and improving its defenses as the election cycle progresses. Decryption She reached out to Anthropic for comment on the findings, but did not immediately receive a response.