OpenAI and Anthropic will establish a war room for the U.S. midterm elections for the first time

According to a report by Semafor, OpenAI and Anthropic will establish dedicated war rooms for the U.S. midterm elections for the first time to ensure that their respective platforms provide accurate election information. An Anthropic spokesperson stated that the company's teams have been strengthening election-related protections and enforcing platform policies throughout the year, and the dedicated team set up before the election will monitor platform abuse and prepare for rapid responses to emergencies.

The practice of tech platforms establishing election war rooms dates back to 2018, when Meta first set up such teams to combat the spread of false information during elections. Today, related risks also include deepfake videos and AI search disseminating misinformation. The report cited that Google AI Mode did not answer the question of who won the 2020 U.S. presidential election last week, and Google stated that it is working on a fix.

The unpredictability of AI outputs, recent cybersecurity issues, and AI search tools' tendency to cater to user expectations all increase the difficulty of monitoring and responding. Katie Harbath, former head of public policy at Meta, stated that the role of the war room is to quickly assess and triage issues according to established procedures when unpredictable problems arise.