Google's AI Mode - the company's AI-generated answer feature built into Search - returned conspiracy theories instead of factual information when a user asked about the death of Food Network chef Anne Burrell.
The query was straightforward: "How did Chef Burrell die?" Instead of pulling from verified news reports, AI Mode assembled a response that included debunked claims and unverified speculation about the circumstances of her death. The kind of answer that, if someone took it at face value, could spread real misinformation about a real person.
This isn't the first time AI-powered search has fumbled sensitive queries. Google has been rolling AI Overviews (formerly called SGE) into more search results throughout 2025 and 2026, and the feature has a pattern of confidently presenting wrong information. Last year, it famously suggested adding glue to pizza and eating rocks for minerals. But there's a meaningful difference between a silly cooking mistake and fabricating details about someone's death.
The core problem is how these systems work. Large language models generate responses by predicting likely next words based on training data. They don't fact-check themselves. When the training data includes a mix of legitimate reporting, forum speculation, and tabloid content, the model has no reliable way to distinguish between them. For unambiguous queries like "what's the capital of France," this works fine. For anything involving contested narratives, recent events, or sensitive personal information, it's a coin flip.
Google does add disclaimers to AI Mode responses, but most users don't read disclaimers. They read the bold text at the top of their search results and move on. That's the whole pitch of AI search - you don't have to click through to sources.
For anyone using AI-powered search for anything beyond basic lookups: verify before you share. Click through to the actual sources listed below the AI summary. The ten seconds it takes to check could save you from repeating something that isn't true.