In February, ETH Zurich and Anthropic published a paper. An entirely automated pipeline analyzed the entire posting history of an anonymous Reddit account—within one minute and for just $2, it identified the real person behind it. No systems were hacked; it simply read posts, searched, and reasoned. The results only went viral on Twitter these past few days. I tried the same approach manually with my own four-year-old Twitter alt account: I pasted all four years of tweets into a large language model’s chat window—and spent nothing. It correctly identified in three sentences that I live in a second-tier city in southern China, own a dog, and tend to post around 8 p.m.—all accurate except one: “worked in cross-border e-commerce.” The location guess was the most precise. 1. Those who anonymously pump coins should be most alarmed. In the past, when a scheme collapsed, you’d delete your account, change your avatar, and restart the next day. Now, a single dialect phrase or a dog’s name mentioned once over four years can be linked by this $2 pipeline to connect multiple accounts to the same person. 2. The same applies to project teams’ anonymous wallets. The withdrawal addresses have always been on-chain, but previously no one bothered piecing them together with tweet timing or linguistic quirks. Now, feed them to a model, and you get results in a minute. I went back and deleted over 300 of my alt account’s tweets containing location references, then fed the cleaned data to the model again—it still figured it out. The $2 in the paper buys you the fully automated version. Your escape alt account? It costs less than a cup of coffee.