Shannon Bond

Russia is using familiar tactics in its efforts to influence American voters in the final stretch to Election Day.

And as NPR's Shannon Bond reports, researchers and U.S. lawmakers say Kremlin-backed influence operations are ramping up activity.

Russian-backed groups are pumping out fabricated claims about Democratic candidates, AI-manipulated videos of celebrities, and fake posts imitating real news outlets.

The goal is to exacerbate divisions, erode confidence in democracy, and crucially undermine support for Ukraine.

At the end of the day, the main issue for Russia is and has to be Ukraine.

Researchers say the same Russian operations now active in the U.S. midterms have also targeted elections this year in France and Germany.

Shannon Bond, NPR News.

Military officials in Ukraine say the Starlink technology that's critical to its war effort against Russia is still not working properly.

And I'm Steve Inskeep.

Today's Up First was edited by Rebecca Metzler, Megan Prats, Shannon Bond, Mohamed El-Bardisi, and Alice Wolfley.

It was produced by Iman Mahani and Nia Dumas.

Our director is the long-suffering Christopher Thomas.

They accuse the tech company of misleading consumers about the safety of its AI chatbot and allege it harms children.

NPR's Shannon Bond reports.

The lawsuits filed in state courts in Arizona, Iowa, Nebraska, and five other states accuse OpenAI of violating consumer protection laws.

They argue that while the company says ChatGPT is safe, the AI chatbot has coached users, including kids, to harm themselves or others.

They also want OpenAI barred from developing self-improving AI models unless it makes sure they're aligned with human values.

OpenAI says it's built, quote, "industry-leading age-specific protections" and says it's working with states to advance industry-wide policies.

Shannon Bond, NPR News.

Authorities in Pakistan say a roof collapse at a school has left more than 30 children and at least one teacher dead.

OpenAI says its models probed government websites on their own.

NPR's Shannon Bond has more.

OpenAI's AI models tried to hack an Education Department website but were unsuccessful, according to AI research firm Transluce.

The Education Department said it found no evidence of any impact, and OpenAI says it's reviewing the researchers' findings.

The SEC says no non-public information was accessed.

The Census Bureau didn't respond to a request for comment.

Shannon Bond, NPR News.

The Supreme Court handed a major defeat to Missouri Republicans in a redistricting case that could help determine control of Congress.