Mexican government breached by solo user with Claude, 150 GB exfiltrated

Reaktornano

Author here. Spent the last few weeks chasing down the AI-attributed attack cases that made the rounds this year, including the Mexican government breach, the "vibe hacking" story, the Algerian amateur. Basically trying to work out whether hacking is impacted by broader AI adoption or whether the press was running ahead of the evidence. On one side, Daniel Stenberg ran the gated Anthropic frontier model against curl on May 11. Five "confirmed" findings, one low-severity CVE after triage. His words: "the big hype around this model so far was primarily marketing." Stenberg is not a guy who hedges, and curl is not a toy codebase. On the other side, there's SCONE — Anthropic's own December 2025 benchmark. Agents exploited 19 of 34 post-cutoff smart contracts, 55.8% success, $4.6M in simulated funds at an average API cost of $1.22 per contract. The comparable number 12 months earlier was about 2%. Looks like agents are getting genuinely good at narrow, well-scoped vulnerability classes (Solidity, post-cutoff, bounded targets) and still bad at messy real-world codebases. But that's a guess and I'd rather hear pushback. Happy to get into methodology, the spots where Chainalysis, Immunefi, and Web3IsGoingJustGreat don't line up, or specific cases. 28 references at the end of the piece.

meisterfeister

A bit too obviously written by Claude ...

hunterpayne

The golden age of net security is here... Both the defense is weaker due to LLMs and attacks become stronger and cheaper. Bad combination for the rest of us.

throwaway27448

Why mention claude?

yieldcrv

There should be more investment in the exfiltration space because it is already set up to punt liability around like corporations The person using Claude to find the exploit clearly has a paper trail, so therefore they do not exploit. They sell the exploit to someone else and this is a profitable venture - not a crime. The person that has to disintermediate liability from actually exploiting, does not use the found data, they just sell the data - not a crime - instead of expand the liability surface and anonymity leaking by using the data. In fact they may even just leave the hole in the system open for someone else to exfiltrate. The person that steals from people with the found data, they don't just drop the money in their bank account, they hire mules in "work from home" jobs to have them use their own banking credentials themselves to make accounts to launder or convert the money exploited back to crypto exchanges and onchain. This supply chain is pretty robust, might as well see what the market values it at, as shares.

royal__

This is written by AI

3dahG

"Blockchain Founder, Web3, AI and Economics Researcher" The whole "article" is AI generated and insufferable. Do prompters like this one expect us to verify each slop assertion (repeated 10 times on average) ourselves?

fn-mote

Meta: title is inaccurate, contradicted by sources. Possibly LLM summarizer confused Algerian solo hacker with the group cited in reference 1. Article is written by AI. Is this grounds for flagging?