Death by Slop: When AI Hallucinations Write Your Obituary

The internet has always been a place where rumors travel fast, but the speed and confidence of modern AI have introduced a new, surreal kind of misinformation. Cliff Stoll—author of the seminal hacking chronicle The Cuckoo's Egg and proprietor of the Klein bottle business—recently found himself in the peculiar position of having to inform people that he is, in fact, still alive.

This scenario unfolded when Stoll began receiving emails from concerned individuals asking if his business was still operating following his passing. The source? An AI-generated review of The Cuckoo's Egg circulating on Facebook. The review didn't just provide the usual synthetic praise; it confidently declared that Stoll had died in May 2024.

The Rise of "AI Slop"

Stoll's experience is a textbook example of what many are now calling "AI slop"—low-quality, AI-generated content designed to flood social media feeds to grow accounts and gain followers. As noted by community members on Hacker News, the barrier to entry for this kind of spam is incredibly low.

"It takes less than a minute to ask an LLM to write a social media post about something interesting and then post it online... the accounts I see seem to be using cheap models that make a lot of mistakes and hallucinate facts."

There is a cynical theory that these hallucinations aren't just bugs, but features. Some suggest that the "extra spice" provided by fabricated facts makes posts feel more interesting, increasing the likelihood that they will be shared by users who haven't yet learned to recognize the hallmarks of synthetic content.

The Performance Delusion

Beyond the technical failure of the LLM, this incident highlights a deeper philosophical shift in how information is produced. Some argue that AI is being developed by people who view everything through the lens of "performance" rather than accuracy. In this framework, the goal is to create a convincing narrative or a compelling piece of content, regardless of whether the underlying facts are true.

This "performance delusion" manifests as a confident lie. The AI does not "know" it is lying; it is simply predicting the next most likely token in a sequence that sounds like a review. When that sequence includes a fabricated death date, the AI presents it with the same authority as a verified fact.

The Feedback Loop of Falsehood

Perhaps the most concerning aspect of this trend is the potential for a recursive loop of misinformation. If AI models are trained on data scraped from the web, and the web is increasingly filled with AI-generated "slop" containing hallucinations, the models may begin to treat these fabrications as ground truth.

This leads to a phenomenon where a hallucination is repeated so often across different platforms that it becomes a "verified" fact in the eyes of subsequent AI models and unsuspecting humans. As one observer noted, this could lead to a future where the internet is viewed as a collection of "artistic works of fiction and falsehood."

The Irony of the Digital Trail

There is a profound irony in this specific instance. Cliff Stoll became famous for his meticulous, human-driven effort to track down a real-world intruder using computer logs in The Cuckoo's Egg. Now, he is being "tracked" and defined by a synthetic entity that creates a digital trail leading to a non-existent grave.

While Stoll handled the situation with a nod to Mark Twain—noting that reports of his death were "slightly exaggerated"—the incident serves as a stark warning. In an era of pervasive AI generation, the burden of verification has shifted entirely to the consumer. When an AI can "kill people off before they notice," the only remaining source of truth is the direct, human confirmation: "I ain't dead yet."