AI chatbots rely on journalism but fails to credit original sources: McGill study finds

A new Canadian study finds popular AI chatbots often rely on journalism to answer questions about current events.

Researchers at McGill University tested around 2,267 Canadian news stories across several leading systems, including ChatGPT, Gemini, Claude and Grok.

They found the tools frequently generated responses based on news coverage without clearly identifying the source around 82 per cent of the time.

The report says most answers failed to credit original reporting, even when details appeared to come directly from news outlets.

The Online News Act requires Meta and Google to compensate media outlets for displaying their content.

Meta pulled news off its platforms in response, but Google has been making payments under the act.

Culture Minister Marc Miller said the government must have a serious conversation about AI systems’ use of news.

With files from The Canadian Press.

WATCH MORE: Meta releases Q1 earnings, adds significant investment in superintelligence