OpenAI Deploys Live AP Vote Counts and Cyber Defense Tools for 2026 Elections

OpenAI is integrating live vote counts from The Associated Press directly into ChatGPT to provide real-time election results for users in the United States and Brazil starting this fall.

Real-time civic data and voting logistics

Users already rely on ChatGPT for practical civic questions regarding registration deadlines and polling locations. To harden the reliability of these responses, OpenAI is partnering with Democracy Works in the US to display verified information about voting and registration processes. This ensures that when users ask about election logistics, they receive data vetted by established civic organizations rather than potentially hallucinated or outdated information.

Globally, the platform is refining web search capabilities to ensure that queries about breaking news or election results are accompanied by direct source links, allowing users to verify information through primary news outlets.

Hardening election infrastructure against cyber threats

Digital infrastructure supporting elections faces increasing pressure from sophisticated cyber actors. OpenAI is positioning its frontier models as defensive assets through its Daybreak initiative. This includes Codex Security, a program designed to automatically identify, validate, and remediate vulnerabilities within developer code.

To support the people managing these systems, OpenAI has extended Codex Security and access to its Trusted Access for Cyber (TAC) program to registered voting system manufacturers in the US. The company is also coordinating with the National Association of Secretaries of State (NASS) and the National Association of State Election Directors (NASED) to brief state authorities on these emerging cyber defense capabilities.

Multi-layered provenance to combat deepfakes

As AI-generated content becomes more prevalent on social media, OpenAI is deploying a dual-layered approach to content provenance. The company has partnered to bring SynthID digital watermarks to images generated via ChatGPT, the OpenAI API, or Codex. These invisible watermarks are designed to persist even after a user takes a screenshot or performs other image transformations.

This watermarking works alongside the C2PA (Coalition for Content Provenance and Authenticity) standard, which uses cryptographic metadata to track an image's history. To bridge the gap between platform-specific tools and the broader web, OpenAI is previewing a public verification tool. This site will allow users to upload off-platform images to detect the presence of SynthID watermarks and surface C2PA metadata, providing a way to verify if an image was created using OpenAI tools.

These technical safeguards aim to provide a signal for social media companies to use when determining how to recommend and distribute civic content in user feeds.

Source: Election information and safeguards in 2026
Domain: openai.com