Anthropic has revealed that Claude blocks just 0.002% of US election-related queries, as it prepares for the 2026 midterms. The company has also introduced new features to provide voters with accurate information.
TL;DR
- Anthropic's Claude blocks only 0.002% of US election-related queries, with most being informational.
- New features provide voters with official voting information and up-to-date election results.
- Anthropic has strengthened its defenses against election-related abuse and misuse.
What happened
Anthropic has shared an update on its election safeguards ahead of the 2026 US midterms, revealing that Claude blocks just 0.002% of US election-related queries. The company analyzed 250,000 conversations over a one-week period and found that only 0.51% were specifically about elections or voting.
Anthropic has introduced two new features to provide voters with accurate information. The first is a banner linking to TurboVote, a nonpartisan resource from Democracy Works, which will appear when users ask about voter registration, polling places, or election dates. The second feature will draw official voting information from Democracy Works into Claude's responses, providing users with state-specific information without leaving the conversation.
Anthropic has also strengthened its defenses against election-related abuse and misuse. The company has deployed new classifiers to better identify novel and multi-step misuse, enhanced its ability to detect election-related misuse by foreign state actors, and built targeted safeguards for localized risks identified in the US, Israel, and Brazil.
Why it matters
For AI/ML developers, this update highlights the importance of building robust safeguards into AI models to prevent misuse and ensure accurate information. It also demonstrates the potential for AI to be used as a tool for political conversations and research.
For startup founders, this update shows the importance of transparency and proactive measures in building trust with users. It also highlights the potential for AI to be used in the political sphere, which could open up new opportunities for startups in this space.
For tech investors, this update underscores the importance of investing in companies that prioritize safety and accuracy in their AI models. It also highlights the potential for AI to be used in the political sphere, which could open up new investment opportunities.
Key facts
- Anthropic analyzed 250,000 conversations over a one-week period and found that only 0.51% were specifically about elections or voting.
- Claude blocks just 0.002% of US election-related queries.
- Anthropic has introduced two new features to provide voters with accurate information, including a banner linking to TurboVote and official voting information from Democracy Works.
- Anthropic has deployed new classifiers to better identify novel and multi-step misuse, enhanced its ability to detect election-related misuse by foreign state actors, and built targeted safeguards for localized risks identified in the US, Israel, and Brazil.
- Anthropic's Claude Opus 5.5, Sonnet 5.5, and Haiku 5.5 models score 99% on the company's political even-handedness evaluation.
- Anthropic has banned accounts involved in election-related abuse, including foreign-based operators of a fake news site and threat actors seeking to exploit the election for financial gain.
Context
Anthropic's update on its election safeguards comes as AI models are increasingly being used in the political sphere. The company's proactive measures to prevent misuse and ensure accurate information highlight the importance of safety and transparency in AI.
The update also underscores the potential for AI to be used as a tool for political conversations and research. As more people turn to AI for information about elections, companies like Anthropic have a responsibility to ensure that the answers they provide are accurate and balanced.
Anthropic's update is part of a broader trend in the AI industry to prioritize safety and accuracy in AI models. As AI continues to evolve, companies will need to invest in robust safeguards to prevent misuse and ensure that AI is used for the benefit of society.
