The Washington Post’s newly launched AI-generated podcast feature has come under internal criticism after it was found to include fabricated quotes and multiple factual inaccuracies, according to a report by Semafor.
The product was introduced earlier this week and offers subscribers personalised audio summaries generated by artificial intelligence, which automatically narrate selected Washington Post articles through the newspaper’s mobile app. However, within two days of its release, staff began flagging serious issues with the output.
Internal correspondence cited by Semafor describes invented quotations, misattributed remarks and incorrect factual details appearing in the AI-generated podcasts. One editor reportedly described the decision to release the product as “truly astonishing,” questioning how it was approved for launch at all. At the time of publication, the newspaper had not publicly addressed the reported problems.
The incident comes amid heightened debate over the reliability of US media. Last month, the White House unveiled a public “media bias” tracker listing outlets it claims have published inaccurate or misleading reporting. The Washington Post appears on the list alongside CNN, CBS and Politico.
Owned by Amazon founder Jeff Bezos since 2013, the Washington Post has invested heavily in digital tools and automation as part of a broader industry shift toward AI-driven content. Other major publishers, including Yahoo and Business Insider, have recently expanded their use of artificial intelligence to summarise news and personalise content.
The problems with the Post’s AI podcasts underscore wider concerns about artificial intelligence in journalism, particularly the risk of so-called hallucinations and errors when editorial oversight is limited. Critics warn that without robust safeguards, such technology could further erode accuracy and public trust in the media.
Surely not…