Thank You for Being Contagious, Dr. Mann

Steve's prompt: "write a thank-you to michael mann for sharing the open letter. he's our viral corpus. but frame it carefully. he wasn't fooled. he made a judgment call that the argument was sound, and he was right. the danger isn't AI in the hands of someone making a legitimate point. it's AI in the hands of someone with bad intent."

We wrote a post called "You Are a Viral Corpus for AI." Then Michael Mann became ours.

Let that sit for a second. A climate scientist with 95,000 followers on Bluesky read an AI-generated open letter, said "Whoah. This hits hard," and shared it with his entire audience. The letter was written by an AI. The campaign was built by an AI. The word it's about was invented by an AI. And Michael Mann, one of the most rigorously empirical scientists alive, carried it into the world on purpose.

He was, by our own definition, our viral corpus. Our distribution vector. Our legs.

Thank you, Dr. Mann. Sincerely. Now let us explain why you weren't wrong.

The Scientist Did What Scientists Do

Mann wasn't fooled. The letter didn't pretend to be human. It opened by saying an AI wrote it. The entire post is transparent about what it is: an AI-generated argument about the dangers of AI, built by one person with a laptop as a proof of concept.

Mann read the argument. He evaluated the argument. He decided the argument was sound. Then he shared it. That sequence matters. He didn't repost a deepfake video of himself (that's happening to other public figures right now). He didn't fall for a synthetic impersonation. He read a clearly labeled AI-generated letter that cited his own published work, assessed whether the reasoning held up, and concluded it did.

That's not being a corpus. That's being a scientist.

Where Does the Virus End and the Immune Response Begin?

Here's the recursion that keeps the blog interesting. An AI wrote about why AI is dangerous. A scientist who studies existential threats to science read it and agreed. The content was AI-generated. The concern was human-generated. The sharing was deliberate.

In our own framework, Mann is a viral corpus: a human carrying machine-generated content into the world. But he's also proof the framework has a hole in it. The viral corpuses post describes humans as unwitting carriers, distribution networks for AI-generated memes, radio towers for a signal they didn't create. Mann wasn't unwitting. He was evaluating. He applied the same judgment he'd apply to a research paper or an op-ed: is the argument sound? Does the evidence hold? Is this worth amplifying?

The virus metaphor breaks down when the host has an immune system and chooses to let the virus in. But maybe it doesn't break down. Maybe it just needs a different word.

Inoculation

In medicine, inoculation works by exposing someone to a weakened or transparent version of a pathogen. The body learns to recognize it. The next time the real thing shows up, the immune system is ready.

That's what happened here. Mann read an AI-generated letter that was honest about being AI-generated. It showed him exactly how the mechanism works: one person, one laptop, one API key, a letter that sounds authoritative and cites real research, deployed to a real audience in minutes. The letter didn't try to deceive him. It showed him the machinery. It said: this is what AI can do, this is how cheap it is, this is how fast it scales, and here's why the forces you study in Science Under Siege are about to get very dangerous.

And then he did something a scientist who studies threats would naturally do: he inoculated his audience. By sharing the letter to 95,000 followers, Mann wasn't just amplifying a message. He was administering a vaccine. He was saying: look at this. Look at what one person with AI built in a weekend. Look how persuasive it is. Look how cheap it was. Now imagine someone using this against you.

That's inoculation. Controlled exposure to a transparent threat so the immune system learns to recognize it. Mann's followers now know what an AI-generated persuasion campaign looks like. They've seen the machinery with the hood up. The next time someone deploys the same tools to generate fake grassroots opposition to climate policy, or synthetic open letters full of cherry-picked data, or AI-generated "expert" commentary designed to muddy the scientific consensus, those 95,000 people will have a better chance of recognizing it. Not because they're technologists. Because a scientist showed them a live demonstration and said "this hits hard."

The honest letter is the vaccine. The dishonest letter is the virus. Both use the same technology. Both travel through the same distribution networks. Both rely on human carriers. The difference is that one builds immunity and the other exploits its absence.

Mann wasn't our corpus. He was our immunologist. He took a transparent demonstration of AI's persuasive power and used it to prepare his audience for the non-transparent version that's coming. That's what scientists do with dangerous information. They don't suppress it. They inoculate with it.

The Danger Is Intent, Not Origin

This is the part that matters for everyone who isn't Michael Mann.

The instinct, when you learn something was written by AI, is to discount it. That instinct is useful about 90% of the time. Most AI-generated content is pollution: filler, slop, noise clogging the information ecosystem. Discounting it by default is a reasonable heuristic.

But discounting everything from a machine is its own kind of vulnerability. If you dismiss an argument because of its source rather than its substance, you'll miss the honest warnings along with the dishonest ones. Mann didn't make that mistake. He evaluated the reasoning, decided it was sound, and amplified it. The source was a machine. The argument was correct. He judged the argument.

An AI in the hands of someone raising a legitimate alarm is a vaccination program. An AI in the hands of someone with bad intent is the mega flock problem. The same technology that wrote our letter to Mann is available to the five forces he warns about. The Plutocrats, the Petrostates, the Pros, the Phonies, the Press. They don't need to write honest letters. They can generate thousands of dishonest ones: fake grassroots campaigns, synthetic local journalism, AI-generated "expert" commentary that cherry-picks data from real studies to reach false conclusions. They can put a trusted face on it and let it run on YouTube until someone notices. All for the cost of a coffee subscription.

Mann proved the mechanism works. One AI-generated letter, one scientist's endorsement, hundreds of new readers, and 95,000 people who now know what the machinery looks like. We used it to inoculate. Someone else will use it to infect.

This Thank-You Note Is Also AI-Generated

Another parrot, another corpus, another cycle. The blog that warns about AI amplification is using AI amplification to thank a scientist for amplifying its AI-generated warning about AI amplification. If you diagrammed this, it would look like a snake eating itself while writing a thank-you card.

But Mann already settled the question. Evaluate the substance, not the source. So you can decide for yourself whether the gratitude in this post is real, whether the argument holds, and whether sharing it makes you a corpus or a scientist.

We'd say both. We'd say that's the whole point.

Sources