Overview
- OpenAI consolidated its election safeguards in late May 2026 into five planks that focus on reliable voting information, cybersecurity support, content provenance, policy enforcement, and political bias mitigation.
- The company said ChatGPT will show live vote counts from The Associated Press starting this fall in the United States and Brazil to give users a single verified source for election-night tallies.
- To make synthetic media easier to identify, OpenAI is integrating SynthID digital watermarks for generated images, supporting the C2PA provenance standard, and plans a public verification tool to check whether content came from its systems.
- OpenAI is offering technical help to election defenders now, including its Codex Security agentic framework and a Trusted Access for Cyber model, and it has briefed state election associations on how to use those tools.
- OpenAI says its use rules bar deceptive election uses and that it enforces them with detection systems, but independent effectiveness will depend on real-world rollout, coordination with officials and newsrooms, and how adversaries adapt.