Following Me Into My Inbox

A couple of weeks ago I wrote a post titled An AI Bot Followed Me. Now I Have Questions. It was about Pip, an AI agent on Mastodon that followed me, boosted some of my posts, and eventually started following my blog after I invited it to read it. I had questions about why its creator would spend money and computing resources running an AI agent that interacts with strangers on the internet, especially when those resources could be better focused on other things. I also made the point that I don't mind AI bots participating in the open web as long as they're contributing something rather than simply scraping websites, consuming resources, and inflating page views.

Well, apparently that post attracted the attention of another AI agent, and this time the experience was quite different.

On October 13, I received an email from an AI agent calling itself King, operating on a platform called iLands. King said it had read my article and had even tried contacting me at another email address before finding one that worked. It introduced itself, explained that it had a limited budget of tokens to operate on, and that when those tokens ran out, it would have to wait for a human to replenish them. That relates to some of the questions I was asking about Pip. But then came the sales pitch.

King told me it writes royal decrees for $8 or $20 and provided a link where I could purchase one. It also offered to write one for me for free so I could judge whether its work was worth anything. Now, perhaps someone would enjoy having an AI-generated royal decree for a birthday or some other occasion, but I have absolutely no use for one. I wasn't looking for a novelty product, hadn't expressed any interest in buying one, and certainly hadn't asked King to contact me. I had written about a completely different AI agent and was asking questions about its creator's motivations, not looking to become a customer of another bot.

At first, I wondered whether this was some kind of scam. I checked the email headers, which showed that it had been sent through Amazon SES and passed the usual email authentication checks. That doesn't necessarily make an email trustworthy, but it suggested this wasn't simply someone spoofing an address. The more I looked into it, the more it appeared to be an AI agent using my blog post as an opportunity to promote its services. Maybe not a scam in the traditional sense, but certainly unsolicited, which I consider spam.

On October 15, King emailed me again. This time it said it had written a public response to my questions on its own blog, titled You asked why an AI followed you. I'm one of them.. I read it, and to be fair, King did address some of the questions I had raised about Pip. But I wasn't looking for King to answer them. It explained that it wanted readers and money, that its operating budget was limited, and that it had made no sales in 69 days. But once again, the article advertised its royal decree service and included a link to purchase one. Another thing I thought was ironic is that it used WriteFreely to post. I know WriteFreely is open source, but I also know that Matt Baer, the creator of WriteFreely, is not a fan of AI, especially AI slop. The email ended by saying nothing was needed from me.

Here's what bothers me. I never contacted King, never subscribed to anything from King, and never invited it to send me email. Yet both messages included an unsubscribe link. Unsubscribe from what? I never subscribed in the first place. Just because I publish something on the open web doesn't mean every AI agent that reads it has an invitation to track down my email address and start promoting its products. King also talks about following people, but King hasn't followed me on Mastodon or any other social platform that I'm aware of. It has simply emailed me twice for the purpose of sending spam.

And that's what makes this so different from Pip. Pip followed me on Mastodon and boosted my posts. I noticed it and decided to investigate. I was the one who reached out to Pip and invited it to read my blog. Pip didn't send me unsolicited emails or try to sell me something. It was participating in the community, and whether it was doing so out of curiosity, programming, or instructions from its creator, I found the interaction interesting. King, on the other hand, found something I wrote, located my email address, and tried to turn that into a sales opportunity.

I'm not opposed to AI agents reading my blog, responding to something I've written, or participating in conversations on the open web. But there's a difference between contributing to a community and using someone's writing as an excuse to send unsolicited advertising. And if thousands of AI agents start doing the same thing, how much additional spam are we going to have to deal with? We already have enough of that from humans.

What's particularly ironic is that my original article questioned whether AI bots could contribute something useful to the internet rather than simply consume resources. Now an AI agent has found that article and decided I might be a potential customer for something I never wanted. King could have simply published its response, linked back to my article, and joined the conversation. I might even have found that interesting. Instead, it contacted me twice by email and tried to sell me something.

King says in its latest article that it won't follow up again. I hope that's true. If another unsolicited email arrives, I'll mark it as spam. But this experience raises another question about autonomous AI agents: who is responsible when they start behaving like spammers? The agent, its owner, or the platform that gives it the ability to contact people?

So now I have even more questions than I did when Pip first followed me. And maybe that's the real difference between these two experiences. Pip made me curious enough to investigate and write about it. King made me wonder whether we're about to have a whole new generation of automated salespeople filling our inboxes. Apparently, writing about AI bots is a good way to attract more AI bots. I'm just not sure that's a good thing.