Margaret Atwood dismisses Claude after spoiler error at Porto festival

Margaret Atwood took aim at Anthropic’s Claude during a packed appearance at the Babell Literary and Cultural Festival in Porto, saying she had used the chatbot exactly once and walked away unimpressed. She said she tested it for a spoiler about the British detective series Father Brown, only to get the wrong answer and, in her view, a lie.

Atwood’s blunt verdict landed in one of the festival’s most sought-after sessions. Organizers said the event had been sold out for weeks before they released 100 additional seats on June 15, underscoring the draw of the author of The Handmaid’s Tale and The Blind Assassin at the first Babell Literary and Cultural Festival.

The festival ran June 24 to 29, 2026, in Porto, Portugal, and Fundação Livraria Lello has promoted it as part of an effort to turn the city into a Book-City. The lineup included major literary figures, including Nobel laureates, giving Atwood’s appearance added weight inside a program meant to position Porto as a literary destination rather than just another stop on the cultural calendar.

Atwood’s criticism sharpened a familiar fault line in public debate over AI accuracy. Claude is marketed by Anthropic as a next-generation AI assistant built to be safe, accurate and secure, but Atwood’s account reduced that promise to a single failed interaction. “The thing about AI is that it’s garbage in, garbage out,” she said, distilling her skepticism into a line that connected a consumer chatbot test to a broader question about what gets reproduced when imperfect inputs scale across publishing, education and public discourse.

That tension matters far beyond one author’s annoyance. In literature, classrooms and newsrooms, AI systems do not generate information in a vacuum. They remix training material, human prompts and editorial choices, which means weak, biased or sloppy inputs can travel quickly when the systems are trusted too far. Atwood’s comments, delivered before an audience that had waited weeks for a seat, turned a pop-culture spoiler into a sharper warning about how easily confidence in machine output can outrun accuracy.