The impact question is really around scale; a few weeks ago Anthropic claimed 500 "high-severity" vulnerabilities discovered by Opus 4.6 (https://red.anthropic.com/2026/zero-days/). There's been some skepticism about whether they are truly high severity, but it's a much larger number than what BigSleep found (~20) and Aardvark hasn't released public numbers.
As someone who founded a company in the space (Semgrep), I really appreciated that the DARPA AIxCC competition required players using LLMs for vulnerability discovery to disclose $cost/vuln and the confusion matrix of false positives along with it. It's clear that LLMs are super valuable for vulnerability discovery, but without that information it's difficult to know which foundation model is really leading.
What we've found is that giving LLM security agents access to good tools (Semgrep, CodeQL, etc.) makes them significantly better esp. when it comes to false positives. We think the future is more "virtual security engineer" agents using tools with humans acting as the appsec manager. Would be very interested to hear from other people on HN who have been trying this approach!
>There's been some skepticism about whether they are truly high severity
To be honest this is an even bigger problem with Semgrep and other SAST tools. Developers just want the .1% of findings that actually lead to issues, but flagging patterns will always lead to huge false positive rates.
I do something similar as what you suggested and it does work well -pattern match + LLM. The downside is this only applies to SAST and so far nobody has found a way to address the findings that make up 90% of a security team's noise, namely SCA and container images.
My first use case of an LLM for security research was feeding Gemini Semgrep scan results of an open source repo. It definitely was a great way to get the LLM to start looking at something, and provide a usable sink + source flow for manual review.
I assumed I was still dealing with lots of false positives from Gemini due to using the free version and not being able to have it memorize the full code base. Either way combining those two tools makes the review process a lot more enjoyable.
> What we've found is that giving LLM security agents access to good tools (Semgrep, CodeQL, etc.) makes them significantly better
100% agree - I spun out an internal tool I've been using to close the loop with website audits (more focus on website sec + perf + seo etc. rather than appsec) in agents and the results so far have been remarkable:
Human written rules with an agent step that dynamically updates config to squash false positives (with verification) and find issues while also allowing the llm to reason.
As a founder of an auditing firm, I definitely feel the heat of the competition when big LLM companies push products that not only compete with us an auditors but also with our own AI-based offerings (https://zkao.io/).
If I were to venture a guess, there's different world in which we might exist in the next 5-10 years.
In one of these futures, we, as auditors, seize to exist. If this is the future, then developers seize to exist too, and most people touching software seize to exist. My guess here is as good as any developer's guess on if their job will remain stable.
In another one of these futures, us auditors become more specialized, more niche, and bring the "human touch" needed or required. Serious companies will want to continue working with some humans, and delegating security to "someone". That someone could be embedded in the company, or they could be a SaaS+human-support system like zkao.
On the other hand, vibe coders will definitely use claude code security, maybe we should call it "vibe security"? I don't mean it as a diss, I vibe code myself, but it will most likely be as good as vibe coding in the sense that you might have to spend time understanding it, it might make a lot of mistakes, and it will be "good enough" for a lot of usecases.
I think that world is a bit more realistic today, than the AGI "all of our jobs are gone in the next years" doom claim. And as
@zksecurityXYZ
, I don't think we're too scared of that world.
These tools have been, and are making us stronger auditors. We're a small, highly specialized team, that's resilient and hard to replace. On the other hand large consultancies and especially consultancies that focus on low hanging fruits like web security and smart contracts are ngmi.
You want frontier models to actively prevent people from using them to do vulnerability research because you're worried bad people will do vulnerability research?
Not at all. I was suggesting if an account is performing source code level request scanning of "numerous" codebases - that it could be an account of interest. A sign of mis-use.
This is different than someones "npm audit" suggesting issues with packages in a build and updating to new revisions. Also different than iterating deeply on source code for a project (eg: nginx web server).
FWIW Claude Code Opus 4.5 ranks ~71% accuracy on the OpenSSF CVE Benchmark that we ran against DeepSource (https://deepsource.com/benchmarks).
We have a different approach, in that we're using SAST as a fast first pass on the code (also helps ground the agent, more effective than just asking the model to "act like a security researcher"). Then, we're using pre-computer static analysis artifacts about the code (like data flow graphs, control flow graphs, dependency graphs, taint sources/sinks) as "data sources" accessible to the agent when the LLM review kicks in. As a result, we're seeing higher accuracy than others.
Haven't gotten access to this new feature yet, but when we do we'd update our benchmarks.
> "Rather than scanning for known patterns, Claude Code Security reads and reasons about your code the way a human security researcher would: understanding how components interact, tracing how data moves through your application, and catching complex vulnerabilities that rule-based tools miss."
Fascinating! Our team has been blending static code analysis and AI for a while and think it's a clever approach for the security use case the Anthropic team's targeting here.
That quote jumped out at me for a different reason... it's simply a falsehood. Claude code is built with an LLM which is a pattern-matching machine. While human researchers undoubtedly do some pattern matching, they also do a whole hell of a lot more than that. It's a ridiculous claim that their tool "reasons about your code the way a human would" because it's clearly wrong--we are not in fact running LLMs in our heads.
If this thing actually does something interesting, they're doing their best to hide that fact behind a steaming curtain of bullshit.
That's a fair point and agreed that human researchers certainly do more than just pattern match. I took it as sort of vision-y fluff and not literally, but do appreciate you calling that out more explicitly as being wrong.
I hope this is better than their competitors products. So far I've been underwhelmed. They basically just find stuff that's already identified by static analysis tooling and toss in a bunch of false positives from the AI scans.
There's a lot of skepticism in the security world about whether AI agents can "think outside the box" enough to replicate or augment senior-level security engineers.
I don't yet have access to Claude Code Security, but I think that line of reasoning misses the point. Maybe even the real benefit.
Just like architectural thinking is still important when developing software with AI, creative security assessments will probably always be a key component of security evaluation.
But you don't need highly paid security engineers to tell you that you forgot to sanitize input, or you're using a vulnerable component, or to identify any of the myriad issues we currently use "dumb" scanners for.
My hope is that tools like this can help automate away the "busywork" of security. We'll see how well it really works.
as a pentester at a Fortune 500: I think you're on the mark with this assessment. Most of our findings (internally) are "best practices"-tier stuff (make sure to use TLS 1.2, cloud config findings from Wiz, occasionally the odd IDOR vuln in an API set, etc.) -- in a purely timeboxed scenario, I'd feel much more confident in an agent's ability to look at a complex system and identify all the 'best practices' kind of stuff vs a human being.
Security teams are expensive and deal with huge streams of data and events on the blue side: seems like human-in-the-loop AI systems are going to be much more effective, especially with the reasoning advances we've seen over the past year or so.
LLMs and particularly Claude are very capable security engineers. My startup builds offensive pentesting agents (so more like red teaming), and if you give it a few hours to churn on an endpoint it will find all sorts of wacky things a human won't bother to check.
I am seeing something closer to the opposite of skepticism among vulnerability researchers. It's not my place to name names, but for every Halvar Flake talking publicly about this stuff, there are 4 more people of similar stature talking privately about it.
People use whatever tools are the most effective and they have plenty of incentive not to talk publicly about them. I think the era of openness has passed us by. But why does stature matter anyway? If I look at chromium or MSRC bug reports, scarcely any of the submitters are from Europe/US and certainly don't have anything resembling stature. That guy hasn't done anything of note in the field in a long time from what I know, he's kind of boomer (you too, no disrespect).
Asking for a friend who’s working on a startup around this general space: do you think it’s better to go niche, focusing on agents for a specific type of application or a specific language/ecosystem, or is that effectively “killing the startup” by limiting market size too soon?
Another question that came up in conversations with them: there might be value in offering a nonscalable, high-touch service, where you build and maintain customized agents tailored to a client’s specific codebase on a periodic basis.
just when European legislators just enshrined SAST scanning into law (Cybersec Resilience Act, Radio Equipment Directive, ...), AI comes around an makes it redundant. Not saying SAST is dead, but sure can't compete with AI today when it's about signal vs. noise.
Is it only crawlers or bots that abuse your product?
We have been developing our own system (1) for several years, and it's built by engineers, not Claude. Take a look — maybe it could be helpful for your case.
Limited preview for researchers, who will be hand picked to write positive reviews.
Enough of this frontier grifting. Make it testable for open source developers at no cost and without login or get lost. You won't of course, because you'd get an unfiltered evaluation instead of guerilla marketing via blog posts, secrecy, and name-dropping researchers that cannot be disclosed.
It’s a free market. The cream will rise to the top eventually regardless of astroturfing or not. And it will be replicated in FOSS too, so no need to be angry.
The impact question is really around scale; a few weeks ago Anthropic claimed 500 "high-severity" vulnerabilities discovered by Opus 4.6 (https://red.anthropic.com/2026/zero-days/). There's been some skepticism about whether they are truly high severity, but it's a much larger number than what BigSleep found (~20) and Aardvark hasn't released public numbers.
As someone who founded a company in the space (Semgrep), I really appreciated that the DARPA AIxCC competition required players using LLMs for vulnerability discovery to disclose $cost/vuln and the confusion matrix of false positives along with it. It's clear that LLMs are super valuable for vulnerability discovery, but without that information it's difficult to know which foundation model is really leading.
What we've found is that giving LLM security agents access to good tools (Semgrep, CodeQL, etc.) makes them significantly better esp. when it comes to false positives. We think the future is more "virtual security engineer" agents using tools with humans acting as the appsec manager. Would be very interested to hear from other people on HN who have been trying this approach!
To be honest this is an even bigger problem with Semgrep and other SAST tools. Developers just want the .1% of findings that actually lead to issues, but flagging patterns will always lead to huge false positive rates.
I do something similar as what you suggested and it does work well -pattern match + LLM. The downside is this only applies to SAST and so far nobody has found a way to address the findings that make up 90% of a security team's noise, namely SCA and container images.
I assumed I was still dealing with lots of false positives from Gemini due to using the free version and not being able to have it memorize the full code base. Either way combining those two tools makes the review process a lot more enjoyable.
100% agree - I spun out an internal tool I've been using to close the loop with website audits (more focus on website sec + perf + seo etc. rather than appsec) in agents and the results so far have been remarkable:
https://squirrelscan.com/
Human written rules with an agent step that dynamically updates config to squash false positives (with verification) and find issues while also allowing the llm to reason.
If I were to venture a guess, there's different world in which we might exist in the next 5-10 years.
In one of these futures, we, as auditors, seize to exist. If this is the future, then developers seize to exist too, and most people touching software seize to exist. My guess here is as good as any developer's guess on if their job will remain stable.
In another one of these futures, us auditors become more specialized, more niche, and bring the "human touch" needed or required. Serious companies will want to continue working with some humans, and delegating security to "someone". That someone could be embedded in the company, or they could be a SaaS+human-support system like zkao.
On the other hand, vibe coders will definitely use claude code security, maybe we should call it "vibe security"? I don't mean it as a diss, I vibe code myself, but it will most likely be as good as vibe coding in the sense that you might have to spend time understanding it, it might make a lot of mistakes, and it will be "good enough" for a lot of usecases.
I think that world is a bit more realistic today, than the AGI "all of our jobs are gone in the next years" doom claim. And as @zksecurityXYZ , I don't think we're too scared of that world.
These tools have been, and are making us stronger auditors. We're a small, highly specialized team, that's resilient and hard to replace. On the other hand large consultancies and especially consultancies that focus on low hanging fruits like web security and smart contracts are ngmi.
Auditors though, they are cooked.
I think you're massively underestimating the complexity and depth of a good security audit service.
Padme: You're scanning for vulnerabilities so you can fix them, Anakin?
Anakin: ...
Padme: You're scanning for vulnerabilities so you can FIX THEM, right, Annie?
but there are open versions available built on the cn OSS models:
https://github.com/lintsinghua/DeepAudit
I hope Anthropic can place alerts for their team to look for accounts with abnormal usage pre-emptively.
This is different than someones "npm audit" suggesting issues with packages in a build and updating to new revisions. Also different than iterating deeply on source code for a project (eg: nginx web server).
We have a different approach, in that we're using SAST as a fast first pass on the code (also helps ground the agent, more effective than just asking the model to "act like a security researcher"). Then, we're using pre-computer static analysis artifacts about the code (like data flow graphs, control flow graphs, dependency graphs, taint sources/sinks) as "data sources" accessible to the agent when the LLM review kicks in. As a result, we're seeing higher accuracy than others.
Haven't gotten access to this new feature yet, but when we do we'd update our benchmarks.
Fascinating! Our team has been blending static code analysis and AI for a while and think it's a clever approach for the security use case the Anthropic team's targeting here.
If this thing actually does something interesting, they're doing their best to hide that fact behind a steaming curtain of bullshit.
I don't yet have access to Claude Code Security, but I think that line of reasoning misses the point. Maybe even the real benefit.
Just like architectural thinking is still important when developing software with AI, creative security assessments will probably always be a key component of security evaluation.
But you don't need highly paid security engineers to tell you that you forgot to sanitize input, or you're using a vulnerable component, or to identify any of the myriad issues we currently use "dumb" scanners for.
My hope is that tools like this can help automate away the "busywork" of security. We'll see how well it really works.
Security teams are expensive and deal with huge streams of data and events on the blue side: seems like human-in-the-loop AI systems are going to be much more effective, especially with the reasoning advances we've seen over the past year or so.
Another question that came up in conversations with them: there might be value in offering a nonscalable, high-touch service, where you build and maintain customized agents tailored to a client’s specific codebase on a periodic basis.
We have been developing our own system (1) for several years, and it's built by engineers, not Claude. Take a look — maybe it could be helpful for your case.
1. https://github.com/tirrenotechnologies/tirreno
Enough of this frontier grifting. Make it testable for open source developers at no cost and without login or get lost. You won't of course, because you'd get an unfiltered evaluation instead of guerilla marketing via blog posts, secrecy, and name-dropping researchers that cannot be disclosed.
No one knows you also caused that problem.