ABRAnews Three sides to every story.

OpenAI and Anthropic open election 'war rooms' to fight AI disinformation in the midterms

OpenAI's team works around the clock, and Google and Meta have also pledged robust efforts to stop their AI tools undermining confidence in elections.

Published 3 min read
Overview

OpenAI and Anthropic have each set up an election "war room" ahead of Tuesday's elections, a first for both companies, according to Semafor. OpenAI's operations center in San Francisco is staffed by more than a dozen people around the clock, and company officials said a satellite center in Washington also takes part. Anthropic has assembled analysts in three cities: San Francisco, New York and Washington.

The New York Times, in a report the Boston Globe ran, says Google and Meta have also pledged strong efforts to keep their AI tools from being used to weaken trust in the democratic process. The Times lists deepfakes, cyberattacks and influence operations, foreign and domestic, among the potential threats. Semafor mentions deepfake videos and AI-powered search results that spread wrong information.

OpenAI has disclosed that Russia, China and Iran tried to use ChatGPT for influence operations, and it says it disrupted two campaigns. It has not yet disclosed a specific midterm effort. Semafor says Meta set up the first tech platform war room in 2018.

The Times says Meta and other platforms later cut their trust and safety teams. Republicans, and after 2025 the Trump administration, had pressed them to, accusing them of censoring conservative views. The Times adds that the administration dismantled government teams that had fought efforts to disrupt voting.

Key figures
  • more than a dozen people staff in OpenAI's San Francisco election operations center — The Boston Globe
  • 1 billion users ChatGPT's user base, as the Boston Globe reports it — The Boston Globe
  • two campaigns foreign influence operations OpenAI says it disrupted — The Boston Globe

Three sides of the story

Liberal3 outlets

AI war rooms arrive after platforms scaled back

The Boston Globe, running a New York Times piece, opens inside OpenAI's San Francisco operations center, where a team of more than a dozen people works around the clock. It quotes Josh Lawson, OpenAI's head of civic and strategic risk: "We’re not just monitoring text anymore." It also quotes Ben Nimmo, OpenAI's chief investigator, on trying to catch operations early.

The piece recalls that Meta and other platforms shrank their safety teams after Republicans, and the Trump administration after 2025, pressed them and accused them of censoring conservative views. It also quotes Lauren Protentis, a former National Security Council official under President Biden, who calls it foolish to assume the threat has dissipated. Semafor calls the war rooms a first for OpenAI and Anthropic, traces the idea to Meta in 2018 and says the AI age has "supercharged" the threats.

It cites Google's AI Mode not answering a 2020 election query and an Anthropic spokesperson on its war room. The New York Times headline says OpenAI and Anthropic set up "war rooms" to fight midterm disinformation, and the outlet says tech companies have vowed to protect Tuesday's elections.

Center3 outlets

AI firms on high alert for Election Day

The Hill and NewsNation report that AI companies, OpenAI and Anthropic among them, are on high alert for Election Day and are following the example of social media firms. The Hill says the industry faces pressure to curb election misinformation, and NewsNation takes readers inside the companies' "war rooms."

Conservative0 outlets

Outlets on this side gave the story little coverage.

All sources: 7 articles from 6 outlets

Liberal (3)

Center (4)

Midterm elections

US elections halfway through a president's four-year term. The whole House, a third of the Senate and many governors are elected. In 2026 Election Day is November 3.

All terms →