Fabricated polls and synthetic respondents: the new threat to data credibility
AAPOR condemned a case of invented polls that reached news outlets and aggregators, and Pew warned that bots, opaque panels and AI-generated surveys complicate election measurement.

On 21 October, the American Association for Public Opinion Research (AAPOR) condemned a new case of fabricated polling. Wedgwood Polling admitted producing fake surveys, which it described as “an experimental project” meant to start a conversation about data integrity and accountability. At least two major news outlets and several aggregators published the numbers before the aggregator FiftyPlusOne grew suspicious, investigated and concluded they were not real results: no one had answered the questions.
AAPOR called data fabrication a serious ethical breach and warned that artificial intelligence makes it easier to produce not only fake data but also fake organisational documentation, making fraudulent pollsters harder to detect. Its president, Mary Losch, warned that bad actors are once again producing and distributing invented data, and urged that this fraud not be confused with legitimate election polling.
A week later, on 29 October, the Pew Research Center published an essay by its president, Michael Dimock, on why the 2026 US midterms are especially hard to poll. He lists the spread of cheap opt-in online panels with opaque methods; bogus respondents, including AI-enabled bot operations and people answering only for payment, which distort estimates for Hispanic and young adults in particular; and “synthetic surveys” that use AI to model what people would say instead of asking them. Pew's research finds the latter are not a reliable substitute for public opinion.
Why it matters beyond the US: the same phenomena, low-cost online panels, bots and language models that simulate answers, are available in any market. Where aggregators are less demanding and rules focus on publication deadlines rather than methods, an invented number can circulate within hours and shape coverage, donations and alliances.
The recommendations converge. AAPOR urges journalists and analysts to vet every pollster, ask about methodology and expect firms to disclose their identity, methods and data under its Code of Ethics. Pew suggests asking who ran the poll, whether it has a track record, whether it has a stake in the result, whether respondents are real people and whether it publishes its questionnaire and method. It also advises against relying on a single poll and to be wary of outliers.
What to watch: whether aggregators tighten their admission filters; whether professional associations move towards mandatory labels for synthetic data; and how online panels respond with quality controls against bots and professional respondents in the final stretch of the cycle.
Before citing a poll in a brief or campaign material, check that the firm exists, has a track record and publishes its method; and if you use synthetic data, present it as modelling, never as a survey.
Summary written by the FOCUS editors based on the original story. Rights to the original content belong to its publisher.
FOCUS courtesy space


