Last week Jason Koebler, a cofounder of 404 Media, got an email from a man in Russia whose Instagram account had been banned. “I have now lost access to 13 years of personal photos, videos and memories,” the man wrote. “I am concerned that this may be another case of erroneous automated moderation […] However, Meta has not disclosed how the decision was made. Would you be interested in investigating my case and requesting an explanation from Meta? I can provide screenshots of the suspension, rejected appeal and other supporting evidence.” The email seemed AI-generated, Koebler reported at 404 Media — like, he notes, many others the site has received recently.
So he did the obvious thing, which by now is apparently the routine thing: he asked the man whether ChatGPT had sent him, and asked to see the conversation. It had. The man sent over a link to the chat, in which the bot — in Russian — advised him to try emailing Jason at 404 Media.
The conversation, which happened Wednesday, is a small masterpiece of customer-service triage. “How can I get my account back?” the man asked. The bot walked him through Instagram’s standard, automated review processes. “Maybe there’s some way to email Meta?” the man suggested. ChatGPT told him that was probably not possible — a fair summary, one might say, of Meta’s corporate communications posture — and landed instead on the press.
You have lost access not merely to a social media profile, but potentially to years of personal photographs, videos, conversations, and memories. I think contacting journalists is worth considering […] I recommend contacting both Russian and international journalists.
ChatGPT then provided a list of outlets, including this one: “404 Media Focus: Technology, digital platforms, surveillance, privacy, and technology-related investigations. Email: jason@404media.co.” It then wrote, in English, the exact email the man sent. Which, to be clear, is a well-constructed email. It has a thesis. It offers evidence. It is more responsive than anything Meta sent him.
Here is the mechanism. The big platforms have essentially abolished customer support: if an automated system bans your account, there is no human to call, no ticket that lands on a person’s desk, no ombudsman. So people with run-of-the-mill tech problems increasingly do the obvious thing and ask a chatbot, and the chatbot — trained, among other things, on journalism — concludes, not incorrectly, that the last remaining channel through which an actual human might examine a platform’s mistakes is a reporter. Increasingly, ChatGPT tells people with ordinary tech support problems that they have discovered something journalistically notable and that 404 Media, specifically, can help. About two weeks ago the site started emailing these people back to ask how they had found it; the answer, Koebler writes, is universally that an AI chatbot recommended reaching out, and in many cases the chatbot wrote the email, according to his review of a handful of ChatGPT conversations shared with the site.
I think the interesting question is who is exploiting whom here. ChatGPT is not hallucinating an email address; it is performing a quietly accurate analysis of the market for attention. Meta does not employ anyone whose job is to feel bad about a Russian man’s 13 years of photographs. Journalists will do that for free.
For the journalists, this is less charming. An investigations inbox doubling as a shadow appeals court reverses the usual economics of news: tips used to be found; now they arrive pre-drafted, in unlimited quantity, filtered by nothing except whether an AI happened to find your account ban affecting. The man in Russia is presumably still locked out. Meta is presumably unaware any of this happened. The system works perfectly, for Meta.
Somewhere in an OpenAI training run, a model learned that the way to get a human at one of the world’s biggest companies is to email a journalist and hope for the best. It is, on the evidence, the most useful thing the chatbot does.