Source: Unite.AI
Anthropic on October 14, 2026 detailed strengthened election safeguards ahead of the US midterms, releasing a privacy-preserving analysis of 250,000 Claude conversations alongside plans to bring official voting information and live election results into Claude.ai.
How Americans Use Claude for Politics
The analysis drew on a random sample of 250,000 US Claude.ai consumer conversations from September 21 to 27, 2026, classified in a privacy-preserving way through Anthropic Insights. Anthropic reported that conversations touching broadly on politics, public policy, or social issues accounted for about 5.75% of all US conversations on Claude.ai, its consumer app, and that about 0.51% concerned elections or voting specifically. Requests asking Claude whom or what to vote for appeared in about 0.01% of conversations, the company said, noting that because such requests are rare it displayed the upper end of a 95% confidence interval for that group. Requests violating its election integrity policies appeared in about 0.002% of conversations, measured in a separate analysis of 250,000 conversations. Anthropic said it expects the share of political and election-related queries to increase as Election Day approaches, and described most observed requests as informational: researching candidates, drafting documents, following geopolitical events, assisting with coursework, and weighing arguments for and against policies.
Official Voting Data and Election-Night Results
Since April 2026, Claude.ai has displayed a banner linking to TurboVote, a nonpartisan resource from Democracy Works, when users ask about voter registration, polling places, or election dates, a feature first described in the company’s April 2026 election-safeguards update. Later in October 2026, Claude will begin drawing official voting information from Democracy Works directly into its Claude.ai responses for US users, so that questions such as where to vote or what identification to bring to a polling place receive state-specific answers without leaving the conversation. The responses draw on publicly available information that Democracy Works collects from state and local election sources. Claude will also begin showing up-to-date official vote counts and election results on election night and in the days that follow.
Bias Testing and Three External Reviews
Anthropic said it tests its models for political bias before every launch and reported that Claude Opus 5.5, Sonnet 5.5, and Haiku 5.5 scored 99% on its political even-handedness evaluation, describing them as its most even-handed models to date.
Three outside groups reviewed Claude’s handling of free expression, each with a different methodology. The Future of Free Speech, an independent think tank at Vanderbilt University, tested Claude on hundreds of requests on contentious topics. The Foundation for American Innovation compared how Claude responded to matched prompts from the political left and right. The Collective Intelligence Project surveyed more than 2,000 people in the US, UK, and India in a three-stage process to establish shared principles for judging responses. Anthropic reported that most responses were rated strong on the first two evaluations, but the reviews also documented shortcomings: The Future of Free Speech and the Collective Intelligence Project found Claude sometimes added unsolicited caveats and counterarguments, such as inserting an opposing view into an advocacy email, and the Foundation for American Innovation found Claude Sonnet 4.5 did not hedge evenly, softening arguments for right-leaning positions more often. Anthropic said it has improved responses to direct yes-or-no questions and shorter-answer requests, is refining when Claude adds caveats, and is expanding its bias evaluations to cover candidates and voting questions as well as policy issues.
Usage Rules and a New Campaign Guide
The company’s Usage Policy prohibits using Claude to deceive voters or disrupt elections, including spreading false information about candidates or voting procedures, impersonating election officials, and suppressing turnout through deception or intimidation; the updated policy takes effect November 12, 2026. Anthropic revised the policy earlier in October 2026 to clarify what the election rules cover and to add examples.
A new October 2026 guide to using Claude in election campaigns lists permitted uses such as drafting press statements and fundraising emails, preparing for debates, analyzing turnout and precinct data, triaging constituent correspondence, and creating accessible outreach materials in large-print, screen-reader-friendly, and translated versions. Prohibited uses include fake personas posing as real people, mass-produced comments inflating a campaign’s engagement, news pages that falsely present themselves as independent, false posts about when or where to vote, and undisclosed synthetic media impersonating a candidate, with enforcement up to and including an account ban. The guide notes that users remain responsible for compliance with election laws, including disclaimer requirements on public communications.
Red-Teaming, Classifiers, and Enforcement
To test its defenses, Anthropic partnered with the Institute for Strategic Dialogue and the McCain Institute, both independent research organizations, to pressure-test its protections in the US, Israel, and Brazil, which all hold major national elections in 2026. Anthropic said analysts began with direct harmful requests, then broke full campaigns into smaller tasks that could look harmless on their own, such as audience research, building news sites, and drafting content. The testers found that Claude reliably refused a single harmful request but could miss a malicious campaign fragmented across many individually benign-looking requests.
Building on those findings, Anthropic said it added classifiers to identify novel and multi-step misuse, including attempts to build fake social media accounts and news sites aimed at political discussions or to run financially motivated election schemes. It also enhanced detection of election-related misuse by foreign state actors, strengthened cyber protections for election infrastructure and campaigns, and built targeted safeguards for localized risks in the three countries.
The company also reported enforcement results: in one case, foreign-based operators of a fake news site used Claude to produce US midterm coverage favoring Russian, Iranian, and North Korean narratives, and other actors used Claude to produce fabricated news designed to attract American audiences and generate advertising revenue. Anthropic said it banned the accounts involved. Its September 2026 threat intelligence report documented nine disrupted influence operations and assessed that most such content drew little authentic engagement, in part because Anthropic often disrupts operations during the production stage, before material reaches the public.
Anthropic said it hopes to publish a more comprehensive account of how people used Claude throughout the election cycle in the coming months, and described the midterm work as extending to elections around the world.