Hacker News Reader

Comment by hyperpape | Hacker News Reader

There's also a tweet by Matthew Prince that says Cloudflare saw thousands of attempts pre-announcement, https://x.com/eastdakota/status/2113079352682570051, then tens of thousands post announcement, and now millions.

Unfortunately, if you read the announcement as referring to 14:00 UTC, that's several hours after Chainguard published their images. I don't know if it's plausible that people saw the Chainguard builds, understood the issue from those and built exploits, but a precise timeline would help understand better.

View on HN

https://x.com/dok2001/status/2112811965735202970 was posted at 7:43 AM UTC on October 21, 2026, and Chainguard (according to the PHP Foundation blog post) shipped at 7:41 UTC. One of the big reasons Chainguard broke the embargo was because Cloudflare had been seeing attacks, IIRC.

The Cloudflare one refers to the time period after the leak where people had already made PoCs. Amazon’s detection is days earlier which is quite concerning.

So either there was another (but non-public) leak or someone else discovered the exploit almost at the same time as it was about to be published. I guess the latter is much less likely because it would be a major coincidence.

The attacker may have known about the exploit for a long time and were saving it, but upon finding out that it was about to be disclosed, they decided it was now or never.

How could he have known on Oct 17 that it was about to be disclosed? The dates only leaked on Oct 20.

You're asking: How could a hacker - with an exploit that potentially allows him/her to break into half the computers on the internet - know some secret information that is stored on a computer on the internet?

It could also just have been a run-of-the-mill exploit in WordPress.

A leak is a much simpler explanation than an independent discovery who then also managed to discover before publication that someone else has also discovered it.

With everybody using LLMs there are no coincidences. Popular LLM platforms are like Oracles that help you identify what other people have been working on. "Find a bug in PHP" might have been useless s few months ago, but once somebody used it to develop a full exploit the LLM will start to regurgitate that information.

> once somebody used it to develop a full exploit the LLM will start to regurgitate that information

Sure about that?

Not my understanding of how the LLMs currently work. There’s a cutoff date and no new info after that date is included in training.

Info to the contrary welcome.

> Not my understanding of how the LLMs currently work. There’s a cutoff date and no new info after that date is included in training.

You are almost certainly correct in terms of how the public models work. I guess it's possible that the labs have some tools connected for more recent stuff like this, but it seems unlikely.

Frontier models absolutely have access to current information. I just asked Claude about my fantasy football line up. It knew current injury reports, line ups and stats. Google search AI had even more information than that.

However, I’ve done some original research regarding an area that the internet has wrong and that doesn’t seem to have back propagated yet.

The models don't have access to new information. The harnesses do. All the major chat interfaces now provide the model with web search tools that they use to locate anything beyond their inherent knowledge (whether it's too new, or just something that didn't get baked in during training).

But if something hasn't been posted online publicly, they won't know about it.

Here they are saying this PHP exploit was found by literally prompting "find bad vulnerabilities in PHP"

https://xxcancel.com/HackingLZ/status/2113025408975143080

Yeah — I have log entries I believe show LLM-bot-driven exploit attempts — POSTs to a variety of non-application URLs, blocked on the front end proxy by rules I added that prevent multipart form POSTs -- from an hour before the patch was released.

And an odd little cluster of potential attempts it caught just after 9am GMT. Again I can't see any reason why I would be getting multipart form posts to those URLs, it's definitely an exploit, but it's possible there's another older exploit with that pattern. They aren't WP-type URLs though.

I got a lot of these as far as my logs go back (which is the 17th also). I don't really think it means anything, just scanning for other vulns, as happens all the time. My server did not get pwned.

https://xxcancel.com/eastdakota/status/2113079352682570051

Reply on news.ycombinator.com