The cost of running a credible disinformation campaign has collapsed. Frances Haugen, the former Facebook data engineer turned whistleblower, puts it plainly: operations that required state-level resources in 2016 can now be executed cheaply with generative AI. The quality ceiling has risen while the price floor has cratered, and content moderation systems — increasingly automated — cannot keep pace with adversaries who simply tweak posts until variants slip through. The structural problem is not that lies exist. It is that the institutional machinery designed to contain them has been deliberately dismantled. After the Cambridge Analytica scandal, social media companies invested heavily in trust and safety teams. A counter-backlash from the right, turbocharged by Elon Musk's acquisition of X and Donald Trump's 2024 election, reversed that investment. Layoffs gutted the relevant teams. Platforms replaced human moderation with AI systems and community-driven tools like Community Notes — cheaper, scalable, and far less effective at catching novel disinformation in non-English languages. Two case studies from the 2026 midterms illustrate the mechanics. In Texas, Democrat James Talarico took time off campaigning with the flu. A rightwing provocateur published a deceptively edited phone call with a hospital staffer, fabricating a claim that Talarico was in an Ohio mental health facility. The Digital Democracy Institute of the Americas tracked how the lie mutated as it crossed into Spanish-language conservative media, morphing from a hospitalization claim into a false report of a suicide attempt. Talarico's campaign released security footage proving he was home in Texas, but the correction could not outrun the mutation. His Republican opponent, Ken Paxton, separately ran AI-generated campaign ads depicting the Presbyterian seminarian supposedly rewriting the Bible. In Pennsylvania, viral posts on X and rightwing blogs claimed an undocumented immigrant arrested in the state was an active canvasser for Democrat Janelle Stelson's congressional campaign. The claim was provably false — the campaign released records showing no paid position — but an official Department of Homeland Security statement amplified it anyway. ICE deputy director Charles Wall issued a public statement treating the fabrication as established fact. Republican congressman Scott Perry continues running ads with the claim. The Stelson campaign sent cease-and-desist letters; a Fox News reporter removed a post; Citizens for Responsibility and Ethics in Washington filed a Hatch Act complaint against DHS. The platforms' own rules are narrow by design. Meta requires disclosure of AI-generated content in election ads and bans misrepresentation of voting logistics, but does little active moderation of political commentary. In 2023, Meta narrowed its restrictions to allow debate around the legitimacy of past elections. YouTube stopped removing content falsely claiming the 2020 election was marred by widespread fraud, citing a chilling effect on political speech. YouTube told the Guardian that all users are subject to the same community guidelines and have access to the same public reporting tools — a statement that is technically true and practically meaningless. The Spanish-language gap is the sharpest vulnerability. Roberta Braga, founder of DDIA, notes that platforms enforce terms of service far more consistently in English than in Spanish. With human moderation teams gutted and AI moderation tools trained predominantly on English-language data, Spanish-language disinformation operates in what amounts to an unmoderated space. This is not a bug in the system — it is the predictable consequence of cost-cutting dressed up as free speech principle. The federal government is no longer a neutral referee. Nora Benavidez of Free Press identifies a pattern: Trump has leveraged agencies including DHS and the US Postal Service for political interference. When an official DHS statement amplifies a fabricated campaign claim against a Democratic candidate, the information environment has not merely degraded — it has been captured. The question facing voters is not whether disinformation exists, but whether any institution retains the authority and will to counter it.
