# Artificial intelligence, from chatbots like Grok to AI deepfakes, is being used as part of a coordinated covert scheme to secretly control or rig the 2026 US midterm elections

**No verdict.** Two different claims sit under this heading, and they have to be held apart. The documented part is real and serious: AI chatbots, Grok among them, have repeatedly given voters false election information, prompting formal warnings from state election officials, and AI deepfakes of candidates have circulated in recent cycles. That much is sourced. The claim we rate is larger and different: that these tools are being wielded as a single, coordinated, covert operation to secretly determine the 2026 midterm outcome. No public evidence establishes such a centralized plot, and no evidence shows AI altering vote counts. This file names no person or company as the author of a scheme, because no proof of one exists; it separates a genuine information-integrity problem from an unestablished conspiracy.

Category: Government & Intelligence · Era: 2020s · First circulated: The underlying worry predates 2026: it built through the 2024 US cycle as AI chatbots were caught giving wrong election information and a deepfake robocall of President Biden hit New Hampshire, and it hardened into a rigging narrative as the 2026 midterm campaign got under way · Believed by: A cross-partisan audience: distrust of AI platforms and their owners runs across the political spectrum, so versions of the claim circulate on both the left (aimed at owners seen as aligned with one party) and the right (aimed at Silicon Valley firms seen as aligned with the other), each pointing at different alleged culprits
URL: https://theconspiratory.com/theory/ai-election-manipulation-2026

## Summary
Going into the 2026 US midterm elections, a claim has spread that artificial intelligence is the hidden instrument of a plot to rig the vote: that AI chatbots such as Grok, together with AI-generated deepfake audio and video of candidates, are being coordinated to covertly control who wins. The claim braids together two things that should be kept separate. One is documented and real: generative AI tools have a well-recorded habit of producing false election information, and synthetic media impersonating candidates has genuinely circulated, both of which election officials and security agencies treat as serious threats to information integrity. The other is the rated conspiracy: that this amounts to a single coordinated covert scheme to secretly predetermine the 2026 result. This file states the record plainly and with sources, then explains why the leap to a centralized rigging operation is, on the public evidence, unproven. It attributes every allegation as an allegation, names no culprit, and accuses no one of a crime.

## The claim
That artificial intelligence, principally chatbots like Grok and AI-generated deepfakes of candidates, is being deployed as part of a coordinated, covert operation, directed by powerful actors who own or control influential AI platforms, to secretly manipulate public opinion and rig or predetermine the outcome of the 2026 US midterm elections, and that this scheme is being concealed from the public.

## Origin and timeline
- 2024-01: On the eve of the New Hampshire primary, thousands of voters receive a robocall using an AI deepfake of President Biden's voice telling them not to vote. It is the first AI voice clone documented to target US voters directly. A political consultant later admits commissioning it, is criminally charged in New Hampshire, and is fined by the FCC.
- 2024-02: Researchers with the AI Democracy Projects and the outlet Proof News report that popular AI models produced inaccurate answers to basic voting questions more than half the time in their testing, framing AI-driven misinformation as a real threat to voters seeking reliable information.
- 2024-08: Secretaries of state from five states write to Elon Musk after X's chatbot Grok tells users, falsely, that Vice President Harris had missed ballot deadlines in nine states. The error was documented to have circulated for days before being corrected. This is a real, sourced instance of an AI chatbot spreading false election information.
- 2024: Post-election reviews (by the Harvard Ash Center, the Knight First Amendment Institute, and others) find that feared AI deepfakes did not, in the end, overwhelm the 2024 cycle: much viral synthetic content was obvious or satirical, and older non-AI 'cheap fakes' were used far more often. The threat is documented as real but its measured impact on outcomes is modest.
- 2025-2026: Grok is increasingly used as an on-platform fact-checker even as studies keep finding it and rival chatbots frequently wrong on political and election questions. Commentary about AI platform owners' partisan leanings sharpens, providing raw material for a stronger claim.
- 2026: As the midterm campaign intensifies, the discrete worries fuse online into a single narrative: that the documented errors and deepfakes are not accidents or isolated bad actors but a coordinated covert scheme to control the 2026 result. Different versions name different alleged culprits depending on the accuser's politics.
- 2026: Fact-checkers and election officials continue to warn voters not to rely on AI chatbots for voting information and to watch for synthetic media, while noting that no evidence supports a centralized plot to rig the vote count. The documented-versus-alleged distinction is repeatedly drawn, and repeatedly collapsed in viral posts.

## The evidence, claim by claim
- Claim: AI chatbots, including Grok, have pushed false election information, which proves they are being used to rig the vote.
  Evidence: The first half is documented; the conclusion does not follow. It is genuinely established that Grok wrongly told users Harris had missed ballot deadlines in nine states in 2024, that five secretaries of state formally objected, and that studies repeatedly find chatbots answering election questions incorrectly. But documented inaccuracy is not documented intent, and misinformation reaching voters is not the same as a scheme to rig an election. Researchers attribute most of these failures to how the models retrieve and generate text (they are built to produce fluent answers, not verified ones), a mechanism that produces error without requiring a plot.
- Claim: AI deepfakes of candidates are circulating, so the election is being secretly manipulated by a coordinated operation.
  Evidence: Deepfakes are real and the concern is legitimate: the 2024 New Hampshire robocall showed a synthetic candidate voice being used to try to suppress votes. But that case points the other way on coordination. It was traced to a single named political consultant who was charged and fined, not to a hidden central operation, and post-election analyses (Knight First Amendment Institute, Harvard Ash Center) found synthetic media was scattered, often crude or satirical, and did not measurably swing results. Scattered bad actors and a unified covert scheme are different claims, and only the first is evidenced.
- Claim: Powerful people own the AI platforms, so they must be steering them to control the election.
  Evidence: Ownership and influence are facts; secret coordinated rigging is an inference the facts do not carry. That a handful of companies and individuals control leading AI systems is true and is itself a reasonable subject of public concern about concentrated power. But motive and opportunity are not evidence of act. A platform giving wrong answers, or an owner holding partisan views, is consistent with ordinary error, bias, and bad design, none of which establishes a concealed operation to predetermine an election outcome. The claim substitutes who could for proof that anyone did.
- Claim: AI is being used to secretly alter or rig the vote count itself.
  Evidence: There is no evidence for this, and it conflates two unrelated systems. The documented AI problems concern information (what chatbots tell voters, what synthetic media depicts), not tabulation. US vote counting runs largely on paper ballots, local administration, audits, and canvassing that the AI tools in question do not touch. No credible reporting has shown AI changing a certified vote total. Sliding from 'AI spreads bad information' to 'AI rigs the count' crosses from a documented problem to an unsupported one.
- Claim: Because different accusers name different culprits, one of them must be right about the plot.
  Evidence: That the same scheme is attributed to opposite actors depending on the accuser's politics is a reason for caution, not confidence. Versions of the claim blame owners aligned with one party and, elsewhere, firms aligned with the other. A genuine coordinated operation would have a specific author and specific evidence; a narrative that reshapes itself to fit whichever side the teller distrusts is behaving like a template applied to anxiety, not like a documented finding. The interchangeability of villains is a hallmark of an unproven story.

## Why people believe it
- The underlying problem is real. Voters have not imagined the errors: Grok did give false ballot-deadline information, chatbots do get election questions wrong at high rates, and a deepfake robocall did target real voters. When the documented threat is genuine, a larger conspiracy built on top of it feels like a natural extension rather than a leap.
- Concentrated power invites suspicion. A small number of companies and individuals really do control the most influential AI systems, and several are publicly enmeshed in politics. It is reasonable to worry about that concentration, and that reasonable worry supplies the emotional foundation the stronger claim rests on.
- AI is opaque. Most people cannot see why a chatbot answered as it did, and that black box makes intentional manipulation and unintentional error look identical from the outside. When you cannot distinguish a bug from a plot, the more alarming reading is easy to reach for.
- It fits the era's distrust of elections. After years of contested claims about how votes are counted, a new and poorly understood technology is a ready vessel for the pre-existing conviction that the process is being manipulated behind the scenes.
- The narrative is cross-partisan and self-sealing. Because it can be aimed at whichever side you already distrust, almost anyone can find a version that confirms their priors, and any AI error that surfaces can be folded in as fresh proof.

## Open questions
- AI-driven election misinformation is a real and unresolved problem. How much documented chatbot error and synthetic media actually changes votes, as opposed to alarming observers, is still being measured, and honest researchers disagree about the magnitude even as they agree the risk is genuine.
- Transparency about how AI platforms are trained, moderated, and instructed is limited, which makes it hard for outsiders to fully rule out deliberate steering in any given instance. Absence of proof of a scheme is not the same as proof of its absence, though the burden remains on the claim to produce evidence.
- The concentration of powerful AI systems in few hands is a legitimate governance concern independent of any rigging claim, and how democracies should regulate that concentration is a live and unsettled question.
- As with any developing story, this file rates the coordinated-rigging claim as of mid-2026 on the public evidence then available. New disclosures could sharpen or change the picture; the documented information-integrity problems, by contrast, are already well established.

## Latest developments
- 2026-07-18T18:21Z: Concrete AI political fakes are now circulating in the 2026 midterm cycle, exactly the documented problem this file separates from the larger rigging claim. In March the National Republican Senatorial Committee released an AI deepfake video of Texas Senate candidate James Talarico (labeled 'AI generated' in small on-screen text) that put fabricated words in his mouth, and a separate attack ad in the Texas primary used a fabricated clip of Sen. John Cornyn appearing to dance with Rep. Jasmine Crockett. These are documented examples of synthetic political media aimed at voters, not evidence of a coordinated covert scheme to control the vote count. (source: https://www.cnn.com/2026/03/13/politics/james-talarico-ai-deepfake-republicans-midterms)

## Sources
- Musk's AI chatbot spread election misinformation, secretaries of state say, Axios (2024): https://www.axios.com/2024/08/05/elon-musk-grok-2024-election-ballot-misinformation
- Secretaries of state urge X to stop its Grok chatbot from spreading election misinformation, TechCrunch (2024): https://techcrunch.com/2024/08/05/secretaries-of-state-urge-x-to-stop-its-grok-chatbot-from-spreading-election-misinformation/
- A political consultant faces charges and fines for Biden deepfake robocalls, NPR (2024): https://www.npr.org/2024/05/23/nx-s1-4977582/fcc-ai-deepfake-robocall-biden-new-hampshire-political-operative
- AI robocalls impersonate President Biden in an apparent attempt to suppress votes in New Hampshire, PBS NewsHour (2024): https://www.pbs.org/newshour/politics/ai-robocalls-impersonate-president-biden-in-an-apparent-attempt-to-suppress-votes-in-new-hampshire
- Risk in Focus: Generative A.I. and the 2024 Election Cycle, Cybersecurity and Infrastructure Security Agency (CISA) (2024): https://www.cisa.gov/resources-tools/resources/risk-focus-generative-ai-and-2024-election-cycle
- AI chatbots are serving up wildly inaccurate election information, new study says, CBS News (2024): https://www.cbsnews.com/news/ai-chatbots-inaccurate-election-information-proof-news/
- AI chatbots got questions about the 2024 election wrong 27% of the time, study finds, NBC News (2024): https://www.nbcnews.com/tech/tech-news/ai-chatbots-got-questions-2024-election-wrong-27-time-study-finds-rcna155640
- Conspiracy and toxicity: X's AI chatbot Grok shares disinformation in replies to political queries, Global Witness (2024): https://globalwitness.org/en/campaigns/digital-threats/conspiracy-and-toxicity-xs-ai-chatbot-grok-shares-disinformation-in-replies-to-political-queries/
- The apocalypse that wasn't: AI was everywhere in 2024's elections, but deepfakes and misinformation were only part of the picture, Harvard Ash Center (2024): https://ash.harvard.edu/articles/the-apocalypse-that-wasnt-ai-was-everywhere-in-2024s-elections-but-deepfakes-and-misinformation-were-only-part-of-the-picture/
- We Looked at 78 Election Deepfakes. Political Misinformation Is Not an AI Problem., Knight First Amendment Institute (2024): https://knightcolumbia.org/blog/we-looked-at-78-election-deepfakes-political-misinformation-is-not-an-ai-problem

Rated by The Conspiratory, a neutral, sourced encyclopedia of conspiracy theories. Full page: https://theconspiratory.com/theory/ai-election-manipulation-2026