Doctors Are Reading the Press Release. Patients Are Reading Each Other.

A family member spent months following medical advice from an AI chatbot — hundreds of dollars in subscriptions, mounting confusion, and eventually a seizure that ended in hospitalization. The post sits on Bluesky without framing or argument, just the sequence of events. It's not a policy paper, but it's doing more work in this conversation than most policy papers will.

That account sits alongside Baltimore emergency dispatch notices stamped "Created with AI, info may be incorrect," and threads unpacking what a 10% diagnostic error rate means when the patient on the receiving end isn't a benchmark dataset. The mood in these conversations isn't politicized outrage — it's the specific dread of people who have already watched something go wrong and are trying to figure out how much of it to trust going forward. Bluesky healthcare threads aren't running hot with the confident alarm of critics who've read the think-pieces; they're running quiet and granular with people who are simply reporting what happened. Legislators are moving to restrict AI in clinical settings, and that news is landing in these communities not as reassurance but as a second grievance — distrust of the technology plus distrust of the people trying to govern it is a harder place to be than distrust of either alone.

The press coverage occupying a different altitude entirely: product launches, partnership announcements, carefully attributed quotes about transformation, clinical trial results from institutions with every incentive to lead with the win. Perplexity's new health product — AI-generated medical summaries with citations — generated exactly the coverage you'd predict: technically credulous, forward-looking, pitched at the reader who wants to believe this is going somewhere good. That framing isn't wrong, exactly, but it's describing a different subject than the one on Bluesky. The coverage environment and the patient environment are tracking the same technology and arriving at incompatible pictures of it.

The asymmetry matters because it determines who shapes the next decision. When the coverage is optimistic and the grassroots conversation is frightened, regulators read the headlines and patients read each other. The result is a policy environment calibrated to the announcement cycle, not to what people with skin in the game are actually experiencing. Perplexity's health product will generate both curves at once — another round of credulous press coverage, another round of anxious personal accounts. One of those curves reaches Capitol Hill. The other reaches the patient in the waiting room. They're not going to converge because they're not describing the same problem.