AI failures · Fabricated content
Perplexity AI — WIRED Finds Inaccurate, Invented Summaries and Covert Scraping; Forbes Alleges Plagiarism
Verified Checked against its sources on 2 sources quoted
- Where it happened
- International or several countries
- Organization
- Perplexity AI
- Date
- June 19, 2024 (WIRED investigation); Forbes demand letter reported June 18, 2024
- Function
- Journalism / Search
- Where it happened
- Answer-engine summarization of news content
- System
- Perplexity AI
- Technology
- Generative AI
- Type of failure
- Fabricated content
- Caught by
- A journalist
Summary
A WIRED investigation published June 19, 2024, found that Perplexity's answer engine at times summarized WIRED stories inaccurately and with minimal attribution, in one case generating text that falsely said WIRED had reported that a named California police officer committed a crime, and in a test invented a story instead of reading the page it was asked to summarize. WIRED also found Perplexity accessing sites that had blocked its crawler. Separately, Forbes accused Perplexity of republishing its reporting, and its general counsel sent a letter demanding removal of the material and repayment of advertising revenue.
Consequence
- Forbes sent Perplexity a letter demanding removal of the infringing articles and reimbursement of advertising revenue earned from them.
Who caught it A journalistFound outside the organization
Analysis
What could have been done
Inputs and source data are complete, current and fit for the use before the AI acts on them.
In a WIRED test the answer engine invented a story instead of reading the page it was asked to summarize, so it produced a summary without the source content in hand.
The system is tested for accuracy, bias and failure modes on representative cases before use.
Inaccurate and invented summaries of real publications, including text falsely saying WIRED had reported that a named police officer committed a crime, were found by WIRED's investigation of the product while it was in public use.
What would likely have prevented it
- Attribute a summary to a named publication only when the system has retrieved that page's text, and return a stated failure, not a summary, when it has not.
- Before release, test summarization on a representative set of articles, including pages the system cannot reach, and check each output against its article for claims the article does not make.
What would likely have caught it earlier
- Sample live answers that name a publication as their source, compare them with the source article, and track invented or misattributed claims as a defect rate with a threshold that triggers escalation.
Key dates
- June 19, 2024Event
- October 16, 2026Checked against sources
- October 18, 2026Checked against sources
- October 21, 2026Checked against sources
Developments are added when the record is reviewed. Next review: October 16, 2027.
Evidence
Each passage is quoted from its source. The quoted passages are re-checked against the live pages every month.
-
In one case, the text it generated falsely claimed that WIRED had reported that a specific police officer in California had committed a crime. (The AP similarly identified an instance of the chatbot attributing fake quotes to real people.)
Supports: inaccurate summaries, false attribution to WIRED, date
-
The letter, dated last Thursday, demands that Perplexity remove the misleading source articles, reimburse Forbes for all advertising revenues Perplexity earned via the infringement
Supports: Forbes demand letter
Suggested citation. Eminence Scan, “Perplexity AI — WIRED Finds Inaccurate, Invented Summaries and Covert Scraping; Forbes Alleges Plagiarism”, Eminence Growth Solutions, last verified October 16, 2026, https://eminencegrowthsolutions.com/tools/eminence-scan/record/fail-039/.
Using our research: you may read and quote Eminence Scan with credit to “Eminence Scan, Eminence Growth Solutions” and a link; all other rights reserved. Terms: Using our research.
Spotted an error, or a development we have missed? Report a correction. How we check records: research methodology.
