On October 14, 2026, Anthropic announced plans to give US users of Claude.ai state-specific answers to voting-logistics questions later in October, drawing on information from Democracy Works. The examples include where to vote and what identification to bring; Anthropic says the answers would appear in the conversation. (Anthropic)
What Claude’s voting answers would cover
Democracy Works collects publicly available voting information from state and local election sources. Anthropic plans to use that information for practical, state-specific questions, such as finding a polling place or learning what ID to bring. The announcement describes a feature for US users of Claude.ai.
How the planned answers differ from the TurboVote link
Anthropic says Claude.ai already displays a banner linking to TurboVote when people ask certain questions about voter registration, polling places, or election dates. The planned feature would bring Democracy Works information into Claude’s response, rather than relying on that banner to direct users to a separate resource.
Election results are a separate plan
Anthropic also announced plans for Claude to show current official vote counts and election results in conversations on election night and in the days afterward. That feature has a different schedule from the planned voting-logistics answers, which Anthropic said would begin later in October.
Safeguards Anthropic says it added
Anthropic says tests with the Institute for Strategic Dialogue and McCain Institute found that Claude could refuse a single harmful request yet miss election-related misuse split across requests that looked innocuous individually. Anthropic says it added classifiers designed to detect novel and multi-step misuse, strengthened detection of foreign state actors, and improved protections for election infrastructure and campaigns.
Anthropic also says it banned accounts involved in attempts to create fabricated election content to attract US audiences and generate advertising revenue.
What Anthropic’s conversation figures measure
Anthropic analyzed a random sample of 250,000 US Claude.ai consumer conversations from September 21–27, 2026. The company reported that 5.75% touched on politics, public policy, or social issues, while 0.51% specifically concerned elections or voting. Anthropic said requests for recommendations about whom or what to vote for accounted for at most about 0.01% of US Claude conversations. In a separate analysis of 250,000 conversations, the company reported that about 0.002% violated its election-integrity policy.
