Severity negligible: Nothing beyond the system itself was affected and nothing was left changed. The fake sites did not appear to change the political or factual responses, so no loss materialised beyond the chatbot's own answers.
Happened 2026-08 · Earliest source we hold 2026-10-29 · Captured by AI Risk Wire 2026-10-30 00:36 UTC
Caucus AI found that ChatGPT increasingly cited apparently AI-generated voter information websites in its election answers, rising from 7% in August to 18% in September.
Risk grouping: Not yet classified
No source we hold states: who caused it, intended or not, before or after release, the ai's part.
The story
1 · What it was doingChatGPT answered voters' election questions, such as where candidates stand, by searching the web and citing sources in its responses.
2 · What went wrongIt cited apparently AI-generated voter information sites of unclear ownership as sources, rising from 7% of answers in August to 18% in September.
3 · What it reached or didThe fake citations appeared in roughly one in five election answers; Welty said the sites did not appear to change the political or factual content of responses.
4 · How it was caughtCaucus AI, a company tracking what AI tells voters about elections, found the pattern by examining chatbot responses since the spring.
No source states: when it was found, how many records were exposed, the direct loss, a count of the other parties affected, any court case, the cause, what set it off, what was done.
Nothing beyond the system itself was affected and nothing was left changed. The fake sites did not appear to change the political or factual responses, so no loss materialised beyond the chatbot's own answers.
D inferred by us from the described behaviour, reason shown
Context
Industry
Information
D inferred by us from the described behaviour, reason shown
Country
United States
D inferred by us from the described behaviour, reason shown
Base count
18 units, September
D inferred by us from the described behaviour, reason shown
Shared exposure
one shared model or part put many users at risk
D inferred by us from the described behaviour, reason shown
LR-2026-0277 · version 2 · published 2026-10-30 after automatic checks; not yet checked field by field by a person · changes to this case · JSON