Generative AI has made it cheap and fast to script, package and spread political narratives, and US adversaries are taking advantage. People familiar with US intelligence and outside researchers say Russia is currently leading online activity tied to the midterms, with China and Iran also active. Officials and researchers are concerned the same AI tactics used overseas could be aimed at US voters for the Nov. 3 contests, although it was not clear that was happening now.
The play is familiar even if the tools are new: sow division and sway opinions. A growing worry is that geopolitical rivals could try to groom chatbots so AI assisted searches elevate misleading sources, which can make false claims look more authoritative. Influence efforts targeting US voters are already under way, according to the people, but there is no evidence of attempts to breach voting systems or alter vote counts.
The campaigns are ramping up even as the Trump administration has shut down US initiatives meant to counter foreign influence - among them the FBI's Foreign Influence Task Force and the Foreign Malign Influence Center housed within the Office of the Director of National Intelligence - arguing those efforts curtailed free speech. "As the capabilities have ramped up, we have ramped down," Bret Schafer said, who holds the senior director post for information operations policy and research at the Institute for Strategic Dialogue. "This feels like the time that we needed all hands on deck working on this new issue."
In an emailed statement, White House spokeswoman Lauren Bis said, "The Trump administration is taking decisive action to protect our elections including prosecuting election crimes, strengthening election infrastructure security, and protecting vulnerable voters from fraud and coercion," President Donald Trump has repeatedly cast aside US intelligence assessments that Russia worked to benefit his 2016 bid, and this summer he used a prime-time address to allege China meddled in US elections in 2020. A Chinese Embassy spokesperson said China is "opposed to the dissemination of disinformation for political purposes and the use of cybersecurity issues to smear China." The Russian Embassy did not reply to an email seeking comment, and the Iranian mission's UN spokesperson was unreachable by phone.
Anthropic reported that Russian- and Iranian-linked operators tried to leverage Claude to assemble webs of bogus social profiles, counterfeit news sites, and additional deceitful material. In September, the company said it terminated accounts that had been using Claude to produce and spread articles for Russian state outlets, including amplifying false claims concerning Moldovan President Maia Sandu in the run-up to the country's 2025 parliamentary elections.
OpenAI and Anthropic say their policies prohibit using their products to deceive people about elections or voting, and both companies say their security teams frequently identify and shut down covert operations and issue public write-ups on them. "Detecting and disrupting adversarial misuse of our models is a core priority for our investigations team," according to an OpenAI spokesperson.
When headlines move quickly it's helpful to get a calm daily read, so sign up for free Market Briefs
Last week, OpenAI reported that a Russian outfit leveraged ChatGPT to assist in creating a sham think tank and to recruit unsuspecting staff, aiming to damage Ukraine's standing in Latin America and, seemingly, to influence political outcomes in the region. The company added that Iranian operators employed AI to improve articles, burnish pitch emails, and craft social media remarks centered on the Iran war and, at times, US politics, and that a dozen online publications around the world carried those materials.
Researchers are seeing chatbots bite too. CASM Technology reported that, in a July report from UK based think tank Demos, Russian operators focused on prompting chatbots rather than traditional search and got them to regard propaganda-group fabrications as legitimate in nearly 500 out of 3,000 queries. "We have evidence that this has been their strategy in Europe," she said.
"Why wouldn't they be doing the same in the US?" The risk rises as usage grows. And because many mainstream outlets block AI scraping on copyright grounds, chatbots can lean more on state media and official sites. A report dated Oct. 12 from the Global Disinformation Index said chatbots may draw propaganda straight from those outlets, particularly when trustworthy material is limited.
People familiar with the intelligence picture say the tactics differ by country. Russian actors tied to the midterms have tried to discredit individual US lawmakers who support Ukraine. China-backed efforts have zeroed in on races connected to US China policy.
Iranian attempts have pushed narratives that undermine confidence in the US election system overall. Measuring impact is tough, in part because many pushes aim to move broader views, like support for Ukraine, rather than a single race. Schafer said the worry now is how easy AI agents make an end to end operation, from reconnaissance to content creation to distribution.
"It's going to make Russia, China and Iran better," he said. The New York Times and the Washington Post first reported on the US intelligence analysis around these foreign influence efforts.
For your money, the practical takeaway is that information shocks can move policy and sentiment. With more convincing AI generated content likely to surface around heated issues, narratives about geopolitics or the health of US democracy can sway debates on defense spending, platform rules and AI regulation, which shape risk and opportunity across sectors.
Staying on top of economic and market noise can be tiring, subscribe to our free Market Briefs newsletter
