OpenAI and Anthropic are establishing dedicated election "war rooms" to monitor their platforms and respond to potential threats during the Nov. 3 U.S. midterm elections, Semafor reported Monday.
The war room concept dates to 2018, when Meta assembled internal teams to monitor election-related misinformation. Eight years later, generative models are amplifying the risks, with the U.S. Election Assistance Commission warning that AI tools "may allow existing threats to scale more quickly and effectively.”
An Anthropic spokesperson told Semafor that cross-company teams are actively monitoring platforms for misuse and coordinating rapid responses to emerging incidents.
The preparations come amid concerns of rising AI-generated disinformation posts flooding social media. Citing experts, The Guardian reported last week that most of such posts are probably “slipping through the cracks” as platforms scale back content moderation.
Separately, Joseph Squillace, a cybersecurity researcher at Penn State, told the university that AI can accelerate existing cyberattacks and make election disinformation cheaper to produce and distribute.
Meanwhile, AI has also given political campaigns another channel to reach voters. Reuters said last week that consultants are helping candidates tailor online content to influence responses from ChatGPT, Claude, and Gemini. About 25% of eligible voters have asked chatbots about political candidates when deciding how to vote, according to the National Conference on Citizenship.