OpenAI Adds Election AI Deepfake Defenses, Protections to Prevent Fraudulent Activity

OpenAI outlined election safeguards for 2026, including ChatGPT access to voting information and live vote counts, tools for checking image provenance, and limits on political persuasion and election-related misuse. The company says its safeguards include partnerships with Democracy Works and the Associated Press and watermarking in collaboration with Google DeepMind’s SynthID.

• OpenAI says its Democracy Works integration provides U.S. users with voting logistics, while an Associated Press partnership supplies live vote-count data for elections in the United States and Brazil.· 1 source
• The company says images generated through ChatGPT, its API and Codex use SynthID watermarks alongside C2PA metadata, and that a public tool at openai.com/verify checks images for those signals.· 1 source
• OpenAI says it prohibits political advertising on its platform and may restrict or terminate access in response to election-related misuse.· 1 source