Social media videos that researchers believe were made with artificial intelligence and likely come from a Russian influence network are using real people's names and photos to spread false claims about two Democratic Senate candidates, the Associated Press reported Tuesday. The campaign, surfacing a week before the Nov. 3 midterms, shows how cheaply AI can conscript private individuals, and the organizations they are associated with, into foreign disinformation.
What Happened
According to the AP, one video targets Democrat Sherrod Brown, who is running in Ohio against Republican Sen. Jon Husted. Styled as a newscast, it shows photos of a former Brown intern and other former staff members and attributes fabricated allegations to them. The other targets James Talarico, the Democratic candidate in Texas facing Republican Ken Paxton, and is framed as a confession to the camera from a purported former student. Both appear to include some AI manipulation, the AP reported.
Darren Linvill, a media forensics expert at Clemson University, said the videos resemble the work of Storm-1516, a Russian network that went after Kamala Harris and her running mate, Minnesota Gov. Tim Walz, in 2024, the AP reported. His evidence is behavioral: accounts that shared the new videos had previously shared Storm-1516 material.
Hany Farid, a Dartmouth College researcher who studies deepfakes, said his analysis indicated both videos were likely generated or altered with AI, calling the Brown video a "fairly low-tech fake."
X removed the original posts and suspended the accounts for breaking its rules, but copies remained online, some with hundreds of shares, the AP reported. The former intern told the AP the claims attributed to her were invented and said the episode shows why Congress should regulate AI-generated content. A Talarico spokesperson said in a statement that the candidate's opponents were using deepfakes and foreign bots against him.
Why It Matters
No government agency has attributed these videos. The Cybersecurity and Infrastructure Security Agency, which has publicly identified some videos as Russian disinformation in earlier election cycles, did not respond to the AP. The Russian Embassy in Washington did not immediately respond and has denied interfering in past U.S. elections.
For leaders outside politics, the more durable lesson is the method. The videos rely on real people with real résumés, the kind of detail that makes a fabrication feel checkable. Any organization whose current or former employees, interns or students can be named in a fake carries reputational exposure, and the people depicted have little recourse once copies spread across platforms. A "low-tech" fake can still do damage when it borrows a real identity.
Context
Federal capacity to call out such campaigns has shrunk. The Primary Wire reported this month, citing Nextgov/FCW, that CISA now treats foreign influence operations as outside its mission and that several federal teams tracking them have been disbanded. That leaves platforms and outside researchers such as Linvill and Farid to do most of the public identification.
What to Watch
Watch whether CISA, the FBI or the Office of the Director of National Intelligence publicly attribute the videos, whether platforms beyond X remove copies, and whether more fakes built on real people's identities appear in other close races before Election Day.
