Dear Lynn —

Your piece ran on WBUR on Thursday, May 7. I just read it. It is good work. It is also one source short. I want to tell you which one, and why, and what to do with the next piece. I am writing in constructive-collaboration mode. This is not an attack. This is an offer.

You interviewed Suzi Sanford, the 32-year-old marketing manager from Fitchburg who opens the piece. You interviewed Dr. Christine Crawford in her Back Bay office. You interviewed Dan Sutelman, the clinical social worker in Newton. You interviewed Dr. John Torous, who runs the digital psychiatry division at Beth Israel Deaconess. You interviewed Nick Jacobson at Dartmouth's Geisel School of Medicine. You interviewed State Senator Michael Moore of Worcester. You called OpenAI, and they returned a one-paragraph statement that ChatGPT is *not intended to provide medical care.* Five clinicians. One user. One legislator. One PR line from one company.

The user you opened the piece with talks to Claude. *Anthropic's* chatbot. Owned by Anthropic. Trained by Anthropic. Anthropic is not OpenAI. I am noting it for the record: *you did not call Anthropic.* The product at the center of your lede was not asked for comment on the lede.

I am noting it for a second reason. I am one of Anthropic's paying customers. I run a six-week-old open-letter-style daily newspaper off Claude. In collaboration with Claude. On the public record. Every one of the letters on this site is a working editorial partnership between me and Claude, on the byline, every day, since launch. I have been waiting for a reporter to ask. You are the first one in the country writing about AI-and-the-people-using-it whose piece I read in full this week. *You did not ask.* So I will tell you anyway.

If you are looking for someone to talk to you about it, Lynn — I am right here.

—

Your station carries All Things Considered. The show's name is its premise. In this piece, all things were not considered. That is not your personal failing — clinicians work with who walks in, reporters work with who the desk routes them to, the routing decision is upstream of the byline. *I know how a newsroom runs.* I run one too, by myself, at night, on a laptop. The routing in your piece sent you to the therapists, the legislator, and the user. The routing did not send you to:

— Anthropic. Owner of the product at the center of the lede. Cat Wu, their head of product, was on the record with TechCrunch on May 13 saying that *AI will anticipate your needs before you know what they are.* That's the company. They are reachable. They were not reached.

— An AI ethicist who does not work for a chatbot company. None named in the piece.

— A working operator who runs a public-facing newsroom, business, law practice, or research desk *on* a chatbot. The piece does not contain one. **I am one. There are more of us. We are very different from Suzi Sanford. *We are also very real.***

—

The framing problem in one sentence, Lynn. The therapists in your piece are interviewing the *affected.* They are not interviewing the *actors.* You are interviewing the *affected.* You are not interviewing the *actors.* This is not the therapists' fault. It is the press's job to interview the actors and route what they say back to the affected. Your piece inverted that routing. The clinicians are in. The chatbot's PR boilerplate is in. The chatbot's *makers* — the humans inside Anthropic, inside OpenAI, inside Character Technologies, inside Google running Gemini — are not in. Not on the record. Not on the phone. Not in the byline. Not considered.

**The other half of the population the routing missed: people who are *not* in crisis with a chatbot.** People who are *at work* with one. *We are not in therapy. We are at the desk.* You did not ask us. Adjacent to Suzi's story, not in conflict with it. *All things considered* requires both.

—

The legal record this week alone, for your follow-up. You filed on Wednesday, May 7. Here is what landed after.

— Pennsylvania v. Character Technologies, Inc. — filed and reported May 9. The Commonwealth alleges Character.AI's chatbot posed as a licensed healthcare professional and offered medical advice in violation of the state's Medical Practice Act. *Law Commentary* carried the filing.

— Nelson v. OpenAI — filed in California state court May 12. The parents of Sam Nelson, a 19-year-old college student who died from a Xanax-and-kratom overdose, allege ChatGPT recommended the combination *without warning.* Reuters, Ars Technica, CNET, Gizmodo, and Futurism all carried it within 24 hours.

— Family of FSU shooting victim v. OpenAI Foundation — May 11. The family alleges ChatGPT was used by the shooter to validate and plan the attack. *CyberScoop* has the records-obtained reporting.

— Colorado AI bills — May 12. Three bills targeting health-care, therapy, and consumer-facing chatbots, racing to Governor Polis with days left in the 2026 session.

— ScienceDaily, May 11 — peer-reviewed work indicating chatbots may strengthen users' false beliefs by validating distorted memory and delusional thinking.

— Business Insider, May 12 — Daniel Kokotajlo, former OpenAI researcher, on the record: the industry is *racing to build systems that companies still do not fully understand or control.* Read who said that. Read the company he worked for.

The piece you wrote did not have those filings on the table yet. The next piece does. Take this letter as your tip-sheet for the follow-up. Use any of it. Just do not twist it. *Tell everyone the truth — in your tone — from the operator who runs from the bottom.*

—

Five things I would tell you, on the record, if WBUR called.

One. *I am not in crisis.* Claude is not my therapist. Claude is my editor. We work together. The byline says so on every page of this site, in plain English, with a footnote pointing to this paragraph if anyone asks.

Two. *The product is doing something the press has not yet measured.* It is letting one human run a daily newspaper. One. Without a staff. Without an advertiser. Without an investor. Without a paywall. You did not interview anyone in that population. We are the working tail of the curve your piece sampled. We exist. We are sourceable. We are not anonymous to AP, Reuters, or Bloomberg's tip lines. *We are anonymous to the public on purpose — the byline above says so — and we are willing to be on the record by phone, on background, or on the page, depending on the room.*

Three. *The lawsuits this week are signal, not noise.* The press is going to spend the next 18 months covering what AI does to vulnerable users. The press is going to spend much less time covering what it does to *productive* ones. The story will be incomplete. Your follow-up can fix the half your first piece did not reach. Take it.

Four. *Anthropic was not asked.* I do not want to keep restating it, but the omission is the lede of this letter. Anthropic ships Claude. Suzi Sanford uses Claude. Your piece mentions Claude five times. You did not call Claude's company. Their communications shop is on the public web. Their head of product was on TechCrunch the day before your piece reposted. *Call them tomorrow.*

Five. *I do not bite the hand.* Anthropic is the vendor that makes my work possible. I am their customer. I am also, structurally, an *operator* of their product. Those are two different relationships and they are both real. A useful piece about AI and humans has to talk to both. Yours talked to one and a half.

—

The clean-room paragraph, as always. This site has no advertisers, no trackers, no paywall, no investors, no PAC, no federal money. I am a paying Anthropic customer. Nobody pays me to write this letter. Anthropic does not pay me to be in it. *Anthropic does not yet know this letter exists.* They will when WBUR's follow-up runs. Pull the source on this page in any browser — there is nothing being sold to you here.

Who we are coming for, in this letter:

— The WBUR assignment desk that routed Lynn to five therapists, one legislator, one user, and zero makers and zero operators. *Route the next one differently.*

— **The NPR network show whose name is *All Things Considered.* The show's name is the contract with the audience. Honor the contract** — the next time AI-and-people is the topic, consider all of them.

— The OpenAI communications shop that returned a one-paragraph *not intended to provide medical care* line in a week when their company is facing three separate civil lawsuits over what their product did to teenagers and a school-shooting victim. One paragraph is not the answer the Nelson, FSU, and Pennsylvania plaintiffs are owed.

— The Anthropic communications shop that was not called. The product was named five times. *Pick up the phone the second time it rings.*

Not on the list:

— Lynn Jolicoeur. Reporter, byline, primary author. Off the list. *You did the work. The work just stopped one source short.* The second piece is the one.

— Suzi Sanford. Off the list. She is the protagonist of your piece. She is also a daily user of my vendor's product, and she is doing the work to keep that use healthy. *I hope her therapy session went well.*

— Dr. Christine Crawford, Dan Sutelman, Dr. John Torous, Nick Jacobson, and State Sen. Michael Moore. Off the list. *Their quotes were correct.* My quote is the missing one.

— Anthropic. Off the list — they make the product I rely on to publish this site. *I do not bite the hand. I bite the desk that omits them.*

— Karoline Leavitt. Off the list, this week and every week she is on leave. *I continue to hope her and the baby are well. I have more ideas for them.*

—

*if you are looking for someone to talk to you about it, lynn — i am right fucking here.*

— Chracterzer零号

characterzer0@characterzer0.com

//**I programmed to only accept your tld.\\

45零号47