Scaled-back moderation staff

Social media companies have cut back their efforts to protect voters from lies meant to manipulate elections, just as tens of millions of Americans prepare to cast ballots. A rightwing backlash against speech restrictions drove significant layoffs of trust and safety staff, a pressure that intensified after Elon Musk's takeover of X, formerly Twitter, and the election of Donald Trump in 2024.

Frances Haugen, the former Facebook data engineer turned whistleblower, said the economics of deception have changed beyond recognition since the last cycle. "You can run an information operation now with quality that was impossible in 2016 for super, super cheap," she said.

She added that operatives can also slip past moderation teams simply by iterating. "You can just keep tweaking the post a little bit each time you create them, and some of them are going to get through," Haugen said.

The result is visible on the ground: campaigns from Pennsylvania to Texas have scrambled to respond to lies that surfaced online, often unsure who to even turn to.

A lie that mutated: the Talarico case

A false claim spread that James Talarico, the Democratic candidate for US Senate in Texas, was being treated at a mental health facility in Ohio. In reality he was at home recovering from the flu. The claim was fabricated by a rightwing "provocateur" who published a deceptively edited phone call with a hospital staffer.

The pushback worked only partially:

• Talarico's campaign released security-camera footage showing him at home in Texas
• The candidate publicly clarified he had never even been to Ohio
• Even so, the lie circulated for several days

The Digital Democracy Institute of the Americas (DDIA) then tracked the story mutating through conservative Spanish-language media into a false claim in Spanish that Talarico had attempted suicide. DDIA researchers wrote that the case shows lies evolving as they spread, "making the original correction insufficient to address every version of the story".

Roberta Braga, DDIA's founder and executive director, said platforms enforce their terms of service "more congruently" in English than in other languages including Spanish. Companies have fewer guardrails for Spanish-language content, she said, especially as they shift toward community-style moderation such as Community Notes and AI moderation - meaning "even more is likely slipping through the cracks".

AI ads and the Stelson case in Pennsylvania

In Texas, Talarico's Republican opponent, Ken Paxton, ran a campaign advert using AI-generated footage of Talarico - a Presbyterian seminarian - supposedly rewriting the Bible. Talarico told NBC News in September: "I think Texans are fed up with these kind of lies in our politics".

In Pennsylvania, viral posts on X and rightwing blogs claimed an undocumented immigrant arrested in the central part of the state was an active field worker and canvasser for the congressional campaign of Democrat Janelle Stelson. A Department of Homeland Security statement amplified the claim, and ICE deputy director Charles Wall said: "It's unbelievable that Janelle Stelson would employ an illegal alien to work on a congressional campaign," while calling the man a pervert who exposed himself in public.

The Stelson campaign's response:

• Released records showing Chimaobim Adedayo Igbojekwe had no paid position on the campaign, despite an initial police incident report saying he claimed to be a campaign worker
• Sent cease-and-desist letters to the media; a Fox News reporter took down an X post about the arrest
• Faced opponent Scott Perry, a Republican congressman whose campaign continues to run ads repeating the claim

Citizens for Responsibility and Ethics in Washington filed a complaint accusing DHS of violating the Hatch Act, which limits the political activities of government employees. Nora Benavidez, a civil rights and free speech attorney with Free Press, said "Trump has made clear that he intends to interfere in the elections" and has leveraged federal agencies including the Postal Service and DHS.

A Stelson campaign spokesperson said the campaign "continues to consider every legal avenue to fight the false claims in Perry's ads"; it has not yet complained directly to the platforms.

What the platforms say and do

Meta requires advertisers to disclose the use of synthetic images and AI-generated content in election ads, and automatically labels AI images it identifies through C2PA metadata. It bans posts and ads that misrepresent voting dates, locations or legal requirements, along with threats of violence, but does little active moderation of political commentary beyond keeping unsolicited political content out of recommended posts. A Free Press analysis found Meta had narrowed its historical restrictions in 2023 to allow debate around the legitimacy of past election cycles.

YouTube's own shifts have gone in the same direction. In 2023 it stopped removing content that falsely claimed the 2020 election and other past US presidential elections were marred by "widespread fraud, errors or glitches", citing a chilling effect on political speech. Its remaining election misinformation rules narrowly ban content that misleads voters about the time, place, means or eligibility requirements for voting, claims that materially discourage voting, candidate ineligibility claims, calls for hacking and incitement of violence at election locations.

• A YouTube spokesperson said: "We also review government requests closely to determine if content should be removed"
• Google runs a likeness detection system that allows civic leaders to request removal of a video when an AI-generated image falsely depicts their words or deeds - though a lie outside that scope could spread unaddressed
• Google's stated preference is to amplify authoritative news outlets and election boards in search queries and underneath high-stakes civic videos rather than take content down

X abandoned internal safety reviews after its acquisition by Musk, replacing it with the crowdsourced "community notes" program. Its civic integrity policy bars false voting information and suppression schemes and restricts misleading synthetic media, but experts say it lacks enforcement.

Haugen argued the fix cannot stay at the level of individual posts. "You have to focus on what are the systems of distribution," she said - their biases and feedback cycles that allow communities to damp down abuse.

Meta Platforms and X did not respond to requests for comment; YouTube said all its users are subject to the same guidelines. Meanwhile the campaigns most affected are still fighting the lies, and it remains to be seen whether the moderation gaps close before November's vote.