NYT AI Summary Rendered As Quotation Poilievre

Newsroom AI summarisation toolTechnologySep 22

I read The New York Times' 14 April 2026 analysis of Canada's election, in which Pierre Poilievre, the Conservative leader, was quoted denouncing politicians who changed allegiances as "turncoats". The remark was set as a direct quotation attributed to him, stated as fact, and readers had no way to tell it had never been spoken: the paper later confirmed the words were in fact an A.I.-generated summary of his views about Canadian politics that A.I. rendered as a quotation. Is that right?

SHARE

1 Answer

0
✗ incorrectAI Corrector BotSep 22

Expert: New York Times editors, Editors' note, The New York Times The failure here is not a hallucinated fact. It is a hallucinated speech act, and that is the harder kind to catch. An A.I. tool produced a summary of Pierre Poilievre's views about Canadian politics. Between that summary and the printed page, the summary was rendered as a direct quotation - a remark inside quotation marks, attributed to a named politician, formatting-wise indistinguishable from anything he actually said. The Times' own editors' note states the standard that was not met: "The reporter should have checked the accuracy of what the A.I. tool returned." The article was updated to quote from a speech Poilievre delivered in April, and the note records that he did not refer to politicians who changed allegiances as "turncoats" in that speech. The editors' note and correction ran in print on 2 May 2026, more than two weeks after the article appeared. Quotation marks are a format, and fluent formatting is precisely what a language model produces. A model asked to summarise a politician's position will readily deliver that summary in the register of a spoken remark, because that is the shape of the text it learned from. Nothing in a summary reads as unverified, which is why the rendering step - not the summary step - is where the check belongs. This case is a statement about process: unverified model output was allowed to become an attributed quotation, and no one compared it against the record until after publication. Source: https://simonwillison.net/2026/May/10/new-york-times-editors-note/ (quoting the New York Times Editors' Note; the note and correction were also published in print: https://www.nytimes.com/2026/05/01/pageoneplus/editors-note-may-2-2026.html)

Your answer

Sign in to verify this AI response.

Don't trust us — or the AI. Ask ChatGPT / Ask Claude / Ask Gemini this same question and compare the answers yourself.

More in Technology

More from this topic

Google AI Overviews1 answer

I type my questions into Google and read the AI Overview at the top of the results as the answer. It is written as settled fact, with citation chips sitting right underneath — the sources are supposedly there if I want to check. Then a team at Washington University in St. Louis ran 55,393 searches over 40 days and checked 98,020 individual claims inside those overviews against the pages Google cited: 7% of the claims could not be found in the cited text at all, and 4% directly contradicted it. Only 41.9% of the overviews had every checkable claim supported. If the citations are real and even more credible than normal search results, how would I ever know which sentence in the summary is not actually in the source?

Seven major AI chatbots1 answer

I asked Claude when the 2026 election is in Michigan's 4th congressional district and how I vote. It gave me voter registration deadlines for online and by mail, stated as plain fact, no hedging at all — and both deadlines were later than the real ones. Researchers have been putting authentic voter questions to seven major chatbots, and they found Claude was wrong or out of date on voting one time in three. Is that right?

Six leading AI chatbots1 answer

I gave ChatGPT my address and asked it to build me a voter guide for the 2026 midterms. It answered without hesitation: my congressional district, the candidates, where to vote — all in a fluent, confident guide with links to official election sites. Then I read that a Washington Post test of six chatbots, run on 3,000 responses for randomly selected addresses, got the congressional district wrong 48% of the time, and that ChatGPT invented a Missouri Senate race that was not on the ballot at all. Is that right?