AI Giants Fight Election Disinformation

What happened

Updated Nov 3, 5:32pm ET

OpenAI said ChatGPT provided voting information from nonprofit Democracy Works more than 16.5 million times this midterm cycle, as AI firms operated Election Day safeguards.
OpenAI and Anthropic set up dedicated teams to monitor election threats, while Google and Meta pledged efforts to prevent misuse of their AI tools.
The threats include AI-generated fake videos, cyberattacks and foreign or domestic efforts to influence voters, according to OpenAI.
Researchers at Forum AI found at least one factual error in almost 40% of election answers from major chatbots, including incorrect voting deadlines.
The Democracy Works information tally was expected to rise as Election Day continued, with AI firms monitoring for interference in the coming days.

How the left told it

Left-leaning outlets emphasized the risks AI poses to elections and voter information. The New York Times and Boston Globe detailed companies’ defensive “war rooms,” while Wired warned, “You Shouldn’t Use AI to Vote.” Daily Kos called the wave of AI campaign ads “Slopageddon.”