AI ads are popular in the 2026 midterms, but state laws haven't stopped a lack of disclosure
Qué dice
The article reports on the use of AI-generated imagery in 2026 campaign advertising, opening with a Bruce Blakeman gubernatorial ad depicting AI versions of Zohran Mamdani and Kathy Hochul without an AI disclosure, which the New York State Democratic Party has complained about to the state Board of Elections. It cites Wesleyan Media Project data identifying at least 164 AI-generated or AI-enhanced political ads this cycle, 69% of which carried no disclosure and 80% of which came from Republican sponsors. Researchers Travis Ridout and Scott Babwah Brennen and strategist Basil Smikle discuss the patchwork of state disclosure laws, enforcement uncertainty, and research showing AI labels can reduce candidate trustworthiness.
Por qué este veredicto
The piece reads as conventional newsroom reporting: named, checkable experts in roles consistent with their real affiliations (Travis Ridout co-directs the Wesleyan Media Project; Scott Babwah Brennen leads NYU's Center on Technology Policy and co-authored an October 2024 report on AI ad labels; Basil Smikle is a known Democratic strategist), a specific dataset with figures and researcher-supplied caveats, a noted request for comment that went unanswered, and cross-referencing against National Conference of State Legislatures data. The scraped text retains Newsday newsletter sign-up and NewsdayTV promo boilerplate, consistent with a capture of a genuine Newsday page rather than synthesized filler; Newsday is an established Long Island daily, so I am overriding the low source_trust prior, which appears to be inferred from the domain name rather than a registry. Specificity signals are high and AI-text markers low. The 2026-specific particulars - Mamdani as NYC mayor, Blakeman as Trump-endorsed gubernatorial nominee against Hochul, Rogers vs. El-Sayed in Michigan, the North Carolina Lee/Bichler ad - postdate my knowledge and are unverifiable to me, but they are internally consistent and plausible extensions of known 2025 dynamics; unfamiliarity is not evidence of falsity. Lack of corroboration in this refresh is expected for a single-outlet enterprise story. No identifiable fabrication, distorted framing, or headline/body mismatch.
Prose is varied, quote-heavy (ten direct quotes attributed to named, real experts), includes hedging and researcher caveats, and carries outlet-specific boilerplate. These are hallmarks of human reporting rather than template-generated filler; automated AI-text markers were low and I agree with that reading.
Afirmaciones examinadas
-
unverifiable
A Blakeman for New York ad used AI-generated images of Mamdani and Hochul without disclosing AI use
Specific, recent ad; described in detail including the 43-second length and 'paid for by' text, but I cannot view it.
-
unverifiable
The New York State Democratic Party filed an August complaint with the state Board of Elections over Blakeman's AI ads
Plausible and specifically attributed to an identifiable filing, but beyond my ability to check.
-
unverifiable
The Wesleyan Media Project identified at least 164 AI-generated or enhanced ads, 69% without disclosure and 80% from Republican sponsors
The project and co-director Travis Ridout are real and routinely publish such ad-tracking releases; the figures come with researcher-stated caveats, which is a mark of genuine sourcing.
-
supported
An October 2024 report co-authored by Scott Babwah Brennen found AI disclaimers can make candidates seem less trustworthy, with a backfire effect
Consistent with published NYU Center on Technology Policy research on AI ad labeling from that period.
-
supported
Tennessee's AI law covers only deepfakes impersonating candidates, and states have varying disclosure requirements
Matches the known patchwork of state AI election laws; the article attributes the cross-reference to NCSL data.
-
unverifiable
Mike Rogers is in a tight Michigan Senate race against Abdul El-Sayed and posted a 'Haters will say it's AI' superhero video
Both are real figures who sought Michigan's open Senate seat; the specific June post is past my knowledge cutoff.
Señales
| Signal | Strength | Score |
|---|---|---|
|
Confianza en la fuente
Estimación previa de reputación del dominio que publica. Un punto de partida, no un veredicto.
|
|
0.28 |
|
Corroboración
Cuántos medios independientes publicaron una noticia coincidente, con descuento cuando todos son de nivel bajo.
|
|
0.14 |
|
Especificidad
Densidad de personas nombradas, citas directas, declaraciones atribuidas, cifras y fechas.
|
|
0.78 |
|
Marcadores de texto de IA
Marcadores estilométricos de relleno generado. Más alto significa más mecánico. Un indicio débil, nunca una prueba.
|
|
0.08 |
|
Carga emocional
Lenguaje sensacionalista, indignado o evasivo. Más alto significa más cargado.
|
|
0.13 |
|
Autoría
Si se acredita a una persona concreta como autora.
|
|
0.00 |
Advertencias
- Nearly all race-specific details postdate my training data and cannot be independently verified; my verdict rests on sourcing structure, internal consistency, and the real identities of the quoted experts.
- The tier-4 source rating appears to be inferred from the domain name; Newsday is an established daily newspaper, and I have weighted the text over that prior.
- No corroborating coverage was found in this refresh, which is expected for single-outlet enterprise reporting and carries little weight either way.
- No independent outlet in this run carried the same story, so corroboration is untested.
Procedencia
| Evaluado por | claude-opus-5 |
| Texto disponible | Cuerpo completo del artículo |
| Encontrado en | Search: campaign ad attack candidate 2026 (discovery canal) |
| Visto por primera vez | Nov 11, 06:05 EST |
| Estimación previa combinada | 0.40 |