Published 9 days ago • loading... • Updated 6 days ago
Spewing US election falsehoods, AI chatbots face scrutiny
In data collected in October, Innovating for the Public Good researchers found 13.9% of chatbot answers to voting-logistics questions were outdated or inaccurate, with 5.9% of errors potentially misleading voters into missing deadlines or voting on the wrong day.
About 42% of American adults now use AI-powered chatbots for political information, though experts warn that generative engine optimization allows third parties to manipulate search results, since "chatbots don't always draw on the best information," said Abdiaziz Ahmed, Brennan Center technology policy strategist.
Across all queries, 5.1% of chatbot answers carried clear factual errors and 10.9% included misleading claims. When facing "attack questions," chatbots shifted to citing sources flagged by Media Bias/Fact Check, an independent media-rating organization.
Meta, Anthropic, OpenAI, and Google stated they are working to prevent election disinformation. Meta spokesperson Cory Chambliss said the company has spent more than $30 billion on safety and security over the last decade, including election protection.
Tim Chambers, primary researcher at the IFPG, called these error rates "extremely troubling," warning that users trusting chatbots could miss voting deadlines or skip required steps, potentially derailing the entire voting process.
Among hyperrealistic videos, an intense tariff war and key states in dispute, Democrats and Republicans are playing Congressional control in a presse where digital truthfulness and the vote of the "packet" will define the political future of the nation.
8 days ago·Caracas, Venezuela (Bolivarian Republic of)