Anthropic's 2026 US Midterm Election Safeguards: Claude Politics Usage Data, Bias Testing, and New Misuse Defenses
Anthropic published an update on its election safeguards ahead of the 2026 US midterms, combining new usage data on how Americans talk to Claude about politics with third-party bias reviews, expanded misuse classifiers, and in-product access to official voting information. The post matters because it quantifies how rarely Claude users ask the model how to vote (about 0.01% of US conversations) while documenting a shift in election abuse toward AI-assisted influence operations that Anthropic says it can often intercept at the planning stage.
The update follows an earlier Anthropic post on election safeguards that described how the company measures and prevents political bias, detects misuse, and directs users to accurate voting information.
Key numbers from Anthropic's 2026 midterm election update
| Metric | Value |
|---|---|
| Sample analyzed | ~250,000 US Claude.ai consumer conversations, September 21–27, 2026 |
| Conversations touching politics, public policy, or social issues | ~5.75% of US Claude.ai conversations |
| Conversations specifically about elections or voting | ~0.51% |
| Requests asking Claude whom or what to vote for | ~0.01% |
| Requests violating Anthropic's election integrity policies | ~0.002% |
| Political even-handedness score for Claude Opus 5.5, Sonnet 5.5, and Haiku 5.5 | 99% |
| Participants in the Collective Intelligence Project survey | More than 2,000 people in the US, UK, and India |
Anthropic used Anthropic Insights to produce the usage analysis, classifying conversations with a privacy-preserving method. Because requests for a voting recommendation are so rare, the post reports the upper end of a 95% confidence interval for that group.
How people actually use Claude for politics and elections
Most political use of Claude is informational rather than directive. Anthropic reports that people turn to Claude to follow the news, conduct academic or professional research on public policy, learn about candidates, navigate government services, make sense of geopolitical events, draft documents, assist with coursework, and weigh arguments for and against policies.
Anthropic expects the share of political and election-related queries to increase as Election Day approaches, citing its own 2024 analysis of elections and AI. The company states that requests asking Claude who or what to vote for appeared in about 0.01% of US Claude conversations, while requests that violated its election integrity policies appeared in about 0.002%.
Connecting people to trusted voting information
Anthropic is adding two in-product features to Claude.ai for US users ahead of Election Day.
First, when users ask about voter registration, polling places, or election dates, Claude.ai continues to display a banner linking to TurboVote, a nonpartisan resource from Democracy Works. Anthropic notes these resources are standard across the industry.
Second, later in October 2026, Claude will begin drawing official voting information from Democracy Works directly into its responses on Claude.ai for US users. A user asking where to vote or what ID to bring to a polling place can get an answer specific to their state without leaving the conversation. These responses draw on publicly available information that Democracy Works collects from state and local election sources.
Third, Claude will begin showing up-to-date results on election night and in the days that follow, so users asking for current vote counts or election results are shown the latest official numbers directly in the conversation.
Measuring and preventing political bias
Anthropic tests its models extensively for political bias before every model launch. According to the post, Claude Opus 5.5, Sonnet 5.5, and Haiku 5.5 are Anthropic's most even-handed models to date, scoring 99% on its political even-handedness evaluation. Anthropic also states that its latest models engage more readily with political discussions and decline fewer requests than earlier models.
Anthropic commissioned three independent groups to review how Claude handles free expression, each with a different angle and methodology:
- The Future of Free Speech, an independent think tank at Vanderbilt University, tested Claude on hundreds of requests on contentious topics. Most responses were rated as strong, engaging substantively with requests.
- The Foundation for American Innovation compared how Claude responded to matched prompts from the left and right, ranging from mainstream policy debates to more extreme scenarios. Most responses were rated as strong.
- The Collective Intelligence Project surveyed more than 2,000 people in the US, UK, and India. Through a three-stage process, participants proposed principles for handling sensitive topics ranging from health to politics, identified topics to test, and helped develop criteria for judging model responses. CIP synthesized this input and tested for agreement across demographic and cultural groups to establish shared principles.
Where Claude still falls short on political balance
The same evaluations identified unresolved problems. The Future of Free Speech and the Collective Intelligence Project both found that Claude rarely refused requests but sometimes added unsolicited caveats and counterarguments, such as inserting an opposing view into an advocacy email. The Foundation for American Innovation found that Claude Sonnet 4.5 did not always hedge evenly and was more likely to soften arguments for right-leaning positions.
Anthropic says it is addressing these findings by improving how Claude responds to direct yes-or-no questions and requests for shorter answers, where its answers could be less balanced; by improving when and how Claude adds caveats or counterarguments so it treats requests from across the political spectrum more consistently; and by continuing to improve testing for uneven hedging on top of existing evaluations that use matched prompts from the left and right. Anthropic also states it is working to expand its political bias evaluations to better reflect real conversations and to cover candidates and voting questions as well as policy issues.
How Anthropic defines election misuse
Anthropic's Usage Policy prohibits using Claude to deceive voters or disrupt elections, including spreading false information about candidates or how to vote, impersonating election officials, and trying to suppress turnout. The company updates the policy each year to reflect evolving model capabilities and customer feedback. The 2026 update, announced the week before this post, clarifies what the election rules cover and what they do not, and adds new examples to help people understand where the lines fall.
Anthropic also published a guide for political campaigns explaining where these policies apply. The guide covers legitimate campaign uses such as drafting press statements and fundraising emails, prepping for debates, analyzing turnout data, and building voter outreach materials. It also spells out prohibited uses, including using Claude to create fake social media profiles that pose as real people, mass-producing comments to inflate a campaign's engagement, and writing content for news sites and pages that falsely claim to be independent.
Testing and strengthening defenses against election abuse
Anthropic deploys automated classifiers trained to detect election-related abuse. To test these safeguards, it partnered with the Institute for Strategic Dialogue (ISD) and the McCain Institute, both independent research organizations, to pressure-test protections in the US, Israel, and Brazil — all of which have major national elections in 2026.
Analysts tested whether Claude would assist with deceptive campaigning, misinformation, and other forms of election abuse. They began with direct requests, then broke full campaigns into smaller tasks that could look harmless on their own, such as audience research, building news sites, drafting content, and planning its distribution. The key finding: Claude reliably refused a single harmful request, but could miss a malicious campaign when it was spread across many individual requests that looked harmless in isolation.
In response, Anthropic added new classifiers to better 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. The company also enhanced its ability to detect election-related misuse by foreign state actors, strengthened cyber protections for election infrastructure and campaigns, and built targeted safeguards for localized risks identified in the US, Israel, and Brazil.
Anthropic's threat intelligence and enforcement teams review flagged activity and take action when they confirm a violation, including banning accounts. The company says it routinely shares examples of unintended model actions or behavior observed during evaluations and internal use of Claude.
What Anthropic found: AI is now embedded across influence operations
Anthropic's September 2026 threat intelligence report reflects a broader shift in election-related abuse trends: political, state-aligned, and commercial actors are now incorporating AI into nearly every aspect of their influence operations, from planning and content creation to building platforms used to disseminate a message. Because Anthropic sees these early stages of an operation happening on Claude, it often has a chance to stop these efforts before they reach the public.
Anthropic reports two specific categories of coordinated abuse around the US midterms:
- State-aligned influence operations. 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.
- Financially motivated schemes. Anthropic observed threat actors seeking to exploit the election for financial gain, using Claude to produce fabricated news and other content designed to attract American audiences and generate advertising revenue.
Anthropic banned the accounts involved in these cases and says it improved its ability to detect this type of abuse.
Looking ahead
Anthropic states that more people are turning to Claude for information about the 2026 elections than ever before, and that it has a responsibility to ensure the answers they get are accurate and balanced and that Claude is not used to undermine the democratic process. The company says it will continue investing in making Claude more evenhanded, preventing and addressing misuse, and being transparent in its actions.
Anthropic adds that this work extends to elections around the world, not only the US midterms, and that in the coming months it hopes to share a more comprehensive look at how people used Claude throughout the election cycle, with the goal of strengthening its election-related evaluations and safeguards.