OpenAI publishes 2026 election safeguards
OpenAI struck a data partnership with the Associated Press to surface live results in ChatGPT and endorsed two federal bills restricting deceptive AI-generated campaign media.
- Government & policy
- Safety & alignment
- Minor
OpenAI set out a package of measures intended to limit election-related misuse of its products ahead of the 2026 US midterms and other elections abroad. The company said ChatGPT would surface live results, sourced from an Associated Press data partnership beginning that autumn in the United States and Brazil, and voting-logistics information supplied by Democracy Works, alongside web-search answers on election questions that link out to their sources.
On misuse, OpenAI pointed to its existing usage policies barring election interference, voter-demobilisation campaigns and deceptive impersonation of candidates or officials, and to its recent adoption of SynthID watermarking on AI-generated images as a tool for distinguishing real from synthetic election imagery. It also said it supported two pieces of federal legislation: the Protect Elections From Deceptive AI Act, which would ban distributing materially deceptive AI-generated audio or video of federal candidates, and the Preparing Election Administrators for AI Act, which would direct the Election Assistance Commission to write voluntary AI guidance for election offices.
The announcement continued a pattern from prior US election cycles in which major AI labs published pre-emptive policy statements on misuse rather than waiting to respond to specific incidents, following comparable moves by Google and Anthropic around the same period.