Meta's chatbot is straight up telling mushroom hunters to cook up poison

Just when it may have seemed like AI couldn’t surprise us anymore, Meta’s “FungiFriend” chatbot showed up in a popular Facebook mushroom foraging group, and soon enough, it advised users to cook a toxic mushroom in butter. Yes, you read that right. It’s a classic example of how AI can go from helpful to dangerous, especially in something as risky as wild mushroom foraging.

According to 404 Media, it’s not even the first time AI has messed up in the mushroom world. Over the past few months, there’s been a surge of AI-written mushroom foraging books on Amazon and AI-generated mushroom images flooding Google searches. Now, Meta has upped the ante by automatically inserting FungiFriend directly into the “Northeast Mushroom Identification & Discussion” group. This group, with its 13,500 members, mostly beginners, is a space where people come for safe foraging advice. The group’s admins are now rushing to remove the bot, while Meta, predictably, remains silent.

FungiFriend’s profile picture? A “bearded, psychedelic wizard,” AI-generated, of course—because if you’re going botch mushroom advice, might as well do it in style. Rick Claypool, a seasoned forager and consumer safety advocate, points out that these bots are nowhere near reliable for identifying edible mushrooms. In Fungi Magazine, he highlighted how these systems lack the skill to distinguish safe mushrooms from dangerous ones.

The risk became real when a group member asked FungiFriend for cooking advice on Sarcosphaera coronaria, a mushroom that soaks up arsenic and has even caused deaths. The bot casually replied that it was “edible but rare,” recommending a sauté or soup. This is as absurd as Google’s AI telling users to put glue on pizza.

Meanwhile, Apple’s new AI Notification Summary isn’t helping AI’s reputation, either. It has frustrated users with summaries that missed critical details or twisted messages out of context, leading to confusion and missed alerts. So, with both Meta and Apple dropping the ball, it’s understandable why people are wary of tech companies pushing AI into every aspect of daily life.

For mushroom foragers, though, this goes beyond annoying—it’s dangerous. Claypool explains that these groups are where people turn for real, human advice on mushroom identification. But with FungiFriend now the top response option on mobile, Meta is pushing people toward AI advice over real experts in the community.

In a typical group post, bad advice can be corrected by others. But an AI? The built-in safety checks are not enough. Beginners, nervous about asking “stupid” questions, may turn to the bot, thinking it’s a judgment-free zone. As Claypool puts it, “The AI might not judge you, but it might get you killed.”