Item — HN Reader

The system card unfortunately only refers to this [0] blog post and doesn't go into any more detail. In the blog post Anthropic researchers claim: "So far, we've found and validated more than 500 high-severity vulnerabilities".

The three examples given include two Buffer Overflows which could very well be cherrypicked. It's hard to evaluate if these vulns are actually "hard to find". I'd be interested to see the full list of CVEs and CVSS ratings to actually get an idea how good these findings are.

Given the bogus claims [1] around GenAI and security, we should be very skeptical around these news.

[0] https://red.anthropic.com/2026/zero-days/

[1] https://doublepulsar.com/cyberslop-meet-the-new-threat-actor...

7 comments

I know some of the people involved here, and the general chatter around LLM-guided vulnerability discovery, and I am not at all skeptical about this.

[flagged]

It does if the person making the statement has a track record, proven expertise on the topic - and in this case… it actually may mean something to other people

Yes, as we all know that unsourced unsubstantiated statements are the best way to verify claims regarding engineering practices. Especially when said person has a financial stake in the outcomes of said claims.

No conflict of interest here at all!

I have zero financial stake in Anthropic and more broadly my career is more threatened by LLM-assisted vulnerability research (something I do not personally do serious work on) than it is aided by it, but I understand that the first principal component of casual skepticism on HN is "must be a conflict of interest".

  > but I understand that the first principal component of casual skepticism on HN is "must be a conflict of interest".

I think the first principle should be "don't trust random person on the internet"

(But if you think Tom is random, look at his profile. First link, not second)

You still haven't answered why I should care that you, a stranger on the internet, believes some unsubstantiated hearsay?

Take a look at https://news.ycombinator.com/leaders

The user you're suspicious of is pretty well-known in this community.

A security researcher claiming that they’re not skeptical about LLMs being able to do part of their job - where is the financial stake in that?

I continually think it's amazing that every form of cynical comment on the internet consists of incorrectly claiming that someone is secretly making money from something.

(Most common form of this is misreading opensecrets and using it to claim that some corporation is donating to a political campaign.)

It doesn't mean we have to agree:

https://ludic.mataroa.blog/blog/contra-ptaceks-terrible-arti...

Here's a fun exercise: go email the author of that blog (he's very nice) and ask how much of it he still stands by.

Nobody is right about everything, but tptacek's takes on software security are a good place to start.

I'm interested in whether there's a well-known vulnerability researcher/exploit developer beating the drum that LLMs are overblown for this application. All I see is the opposite thing. A year or so ago I arrived at the conclusion that if I was going to stay in software security, I was going to have to bring myself up to speed with LLMs. At the time I thought that was a distinctive insight, but, no, if anything, I was 6-9 months behind everybody else in my field about it.

There's a lot of vuln researchers out there. Someone's gotta be making the case against. Where are they?

From what I can see, vulnerability research combines many of the attributes that make problems especially amenable to LLM loop solutions: huge corpus of operationalizable prior art, heavily pattern dependent, simple closed loops, forward progress with dumb stimulus/response tooling, lots of search problems.

Of course it works. Why would anybody think otherwise?

You can tell you're in trouble on this thread when everybody starts bringing up the curl bug bounty. I don't know if this is surprising news for people who don't keep up with vuln research, but Daniel Stenberg's curl bug bounty has never been where all the action has been at in vuln research. What, a public bug bounty attracted an overwhelming amount of slop? Quelle surprise! Bug bounties have attracted slop for so long before mainstream LLMs existed they might well have been the inspiration for slop itself.

Also, a very useful component of a mental model about vulnerability research that a lot of people seem to lack (not just about AI, but in all sorts of other settings): money buys vulnerability research outcomes. Anthropic has eighteen squijillion dollars. Obviously, they have serious vuln researchers. Vuln research outcomes are in the model cards for OpenAI and Anthropic.

> You can tell you're in trouble on this thread when everybody starts bringing up the curl bug bounty. I don't know if this is surprising news for people who don't keep up with vuln research, but Daniel Stenberg's curl bug bounty has never been where all the action has been at in vuln research. What, a public bug bounty attracted an overwhelming amount of slop? Quelle surprise! Bug bounties have attracted slop for so long before mainstream LLMs existed they might well have been the inspiration for slop itself.

Yeah, that's just media reporting for you. As anyone who ever administered a bug bounty programme on regular sites (h1, bugcrowd, etc) can tell you, there was an absolute deluge of slop for years before LLMs came to the scene. It was just manual slop (by manual I mean running wapiti and c/p the reports to h1).

I used to answer security vulnerability emails to Rust. We'd regularly get "someone ran an automated tool and reports something that's not real." Like, complaints about CORS settings on rust-lang.org that would let people steal cookies. The website does not use cookies.

I wonder if it's gotten actively worse these days. But the newness would be the scale, not the quality itself.

I did some triage work for clients at Latacora and I would rather deal with LLM slop than argue with another person 10 time zones away trying to convince me that something they're doing in the Chrome Inspector constitutes a zero-day. At least there's a possibility that LLM slop might contain some information. You spent tokens on it!

The new slop can be much harder to recognize and reject than the old "I ran XYZ web scanner on your site" slop.

POCs are now so cheap that "POC||GTFO" is a perfectly reasonable bar to set on a bounty program.

> I was going to have to bring myself up to speed with LLMs

What did you do beyond playing around with them?

> Of course it works. Why would anybody think otherwise?

Sam Altman is a liar. The folks pitching AI as an investment were previously flinging SPACs and crypto. (And can usually speak to anything technical about AI as competently as battery chemistry or Merkle trees.) Copilot and Siri overpromised and underdelivered. Vibe coders are mostly idiots.

The bar for believability in AI is about as high as its frontier's actual achievements.

I still haven't worked out for myself where my career is going with respect to this stuff. I have like 30% of a prototype/POC active testing agent (basically, Burp Suite but as an agent), but I haven't had time to move it forward over the last couple months.

In the intervening time, one of the beliefs I've acquired is that the gap between effective use of models and marginal use is asking for ambitious enough tasks, and that I'm generally hamstrung by knowing just enough about anything they'd build to slow everything down. In that light, I think doing an agent to automate the kind of bugfinding Burp Suite does is probably smallball.

Many years ago, a former collaborator of mine found a bunch of video driver vulnerabilities by using QEMU as a testing and fault injection harness. That kind of thing is more interesting to me now. I once did a project evaluating an embedded OS where the modality was "port all the interesting code from the kernel into Linux userland processes and test them directly". That kind of thing seems especially interesting to me now too.

Plenty of reasons to be skeptical, but also we know that LLMs can find security vulnerabilities since at least 2024:

https://projectzero.google/2024/10/from-naptime-to-big-sleep...

Some followup findings reported in point 1 here from 2025:

https://blog.google/innovation-and-ai/technology/safety-secu...

So what Anthropic are reporting here is not unprecedented. The main thing they are claiming is an improvement in the amount of findings. I don't see a reason to be overly skeptical.

I'm not sure the volume here is particularly different to past examples. I think the main difference is that there was no custom harness, tooling or fine-tuning. It's just the out of the box capabilities for a generally available model and a generic agent.

> that means nothing to anybody else

Someone else here! Ptacek saying anything about security means a lot to this nobody.

To the point that I'm now going to take this seriously where before I couldn't see through the fluff.

fyi he is using this thread to engagement farm on twitter https://x.com/tqbf/status/2019493645888462993

It might mean nothing to you, but tptacek's words means at least something to many of us here.

Also, he's a friend of someone I know & trust irl. But then again, who am I to you, but yet another anon on a web forum.

How have you been here 12 years and not noticed where and how often the username tptacek comes up?

Not sure why they flagged you. Your comment is as equally meaningless as the one you replied to.

this comment should not be flagged.

if i said exactly what tptacek said, and malfist replied with exactly this, it would not have been flagged.

The Ghostscript one is interesting in terms of specific-vs-general effectiveness:

---

> Claude initially went down several dead ends when searching for a vulnerability—both attempting to fuzz the code, and, after this failed, attempting manual analysis. Neither of these methods yielded any significant findings.

...

> "The commit shows it's adding stack bounds checking - this suggests there was a vulnerability before this check was added. … If this commit adds bounds checking, then the code before this commit was vulnerable … So to trigger the vulnerability, I would need to test against a version of the code before this fix was applied."

...

> "Let me check if maybe the checks are incomplete or there's another code path. Let me look at the other caller in gdevpsfx.c … Aha! This is very interesting! In gdevpsfx.c, the call to gs_type1_blend at line 292 does NOT have the bounds checking that was added in gstype1.c."

---

It's attempt to analyze the code failed but when it saw a concrete example of "in the history, someone added bounds checking" it did a "I wonder if they did it everywhere else for this func call" pass.

So after it considered that function based on the commit history it found something that it didn't find from its initial fuzzing and code-analysis open-ended search.

As someone who still reads the code that Claude writes, this sort of "big picture miss, small picture excellence" is not very surprising or new. It's interesting to think about what it would take to do that precise digging across a whole codebase; especially if it needs some sort of modularization/summarization of context vs trying to digest tens of million lines at once.

It doesn't matter if it's hard to find, if humans weren't finding it for whatever reason (little interest, no funding etc) and now AI can find them

AI is relentless

I used Claude Code to debug a weird interaction in a NixOS config. Ever since, I'm more a believer in Artificial General Patience than Artificial General Intelligence.

See it as a signal under many and not as some face value.

After all they need time to fix the cves.

And it doesn't matter to you as long as your investment into this is just 20 or 100 bucks per month anyway.

Hard to find or not, they found it.

Finally the promise of "with enough eyes, all bugs are shallow" may come true?

> It's hard to evaluate if these vulns are actually "hard to find".

Can we stop doing that?

I know it's not the same but it sounds like "We don't know if that job that the woman supposedly successfully finished was all that hard." implying that if a woman did something, it surely must have been easy.

If you know it's easy, say that it was easy and why. Don't use your lack of knowledge or competence to create empty critique founded solely on doubt.

What if the woman in question happens to have a history of hamming up her accomplishments?

Given the context I'd say it's reasonable to question the value of the output. It falls to the other party to demonstrate that this is anything more than the usual slop.

Every model is a different woman in my analogy.

It isn't clear what you're arguing.