Twitxy Saved

Xint

Verified account This account is verified. Learn more

@xint_official

What if the world's best hackers rebuilt AppSec from the ground up with AI?

Meet Xint - autonomous, comprehensive, fast, and actionable.
xint.io Joined February 2026
33 Following 2.3K Followers 401 Posts

Latest posts 5

Pinned post

Xint Verified account @xint_official
🚨 Critical nginx vulnerabilities disclosed. Immediate action recommended.

If you run nginx, check whether your servers are affected and update to the patched version immediately. If you cannot upgrade, follow the mitigation guidance at nginx2root.com

Today we’re disclosing nginx2root, five distinct vulnerabilities in one of the world’s most widely deployed web servers, each capable of unauthenticated remote code execution under affected conditions. One of these vulnerabilities affects nginx’s most common configuration. We’ve also demonstrated a fully remote, unauthenticated exploit chain that escalates from nginx RCE to root access on the underlying server, using a separate Linux kernel vulnerability.

Why this is so important:
One of these bugs has been hiding in nginx’s HTTP/2 implementation since 2015. That’s more than a decade in software that powers roughly a third of the web. Every official release from 1.9.5 through 1.31.8 contains the vulnerable code, including the master branch as of Nov 10, 2026.

A Xint researcher found the specific vuln that hits nginx in its default config. This discovery was around the same time he found another critical vulnerability in a different project that we cannot talk about yet due to the same concerns as nginx2root. We took those learnings and incorporated them into the harness as a new, general Xint feature (that is, not fine-tuned to nginx), which then found the other 4 critical nginx vulnerabilities with no human guidance.

We coordinated disclosure with the nginx maintainers and major cloud providers to help make fixes available before publication.

Technical details, affected versions, and remediation guidance:
nginx2root.com
4 replies 87 reposts 193 likes 29K views View on X
Xint Verified account @xint_official
@lconstantin at @CSOonline did the hard work of reading through all the blogs of the different researchers credited with finding the critical nginx vulnerabilities announced Nov 18. Great piece for understanding the differences between the 5 distinct vulnerabilities and their individual risk profiles
csoonline.com/article/425046…
www.csoonline.com AI unearths ‘internet breaker’ flaws in Nginx web server The vulnerabilities — including a 10-year-old RCE flaw — add to the recent spate of AI-discovered cracks in foundational internet technologies.
Xint Verified account @xint_official
"Disclose before every distro has a patch, and you're irresponsible. Wait longer, and you're accused of leaving defenders in the dark, without enough time to prepare for the storm.

Release a PoC that doesn't reliably work in real-world environments, and people call it overhyped. Release the fully weaponized, reliable exploit we actually have, and those same people would probably call us reckless...we deliberately choose NOT to release our most advanced, weaponized exploits.

We have exploits that work reliably in real-world conditions, even with unpredictable network traffic and other interference...Releasing those exploits would meaningfully lower the barrier to exploitation and potentially cause real damage."
Brian Pak Verified account @brian_pak 12h
You can't make everyone happy with vulnerability disclosure.

Disclose before every distro has a patch, and you're irresponsible. Wait longer, and you're accused of leaving defenders in the dark, without enough time to prepare for the storm.

Release a PoC that doesn't reliably work in real-world environments, and people call it overhyped. Release the fully weaponized, reliable exploit we actually have, and those same people would probably call us reckless.

I get it. There's no perfect answer.

But here's where we stand.

We care deeply about the security industry and the ecosystem we're part of. We want to push the boundaries of what's possible, contribute meaningful research, and do it responsibly.

And let's be clear about something else.

Of course we want recognition for the work we do. Of course we want the biggest and most respected VCs in the world to invest in us. Of course we want the biggest companies in the world as our customers.

We're building a company. We're ambitious. We're not going to pretend otherwise.

And none of that happens if nobody knows who we are or what we're capable of. So yes, we're going to make some noise. We're going to talk about our work, celebrate our achievements, and take every opportunity to show the world what we're building.

Of course we're going to market the hell out of genuinely cool work. Why wouldn't we?

And yes, we're really fucking good at computers.

But there's a line between promoting your work and putting the community at risk for publicity. We have no intention of crossing it.

That's why we make deliberate choices about what we disclose and what we don't.

We choose to disclose our findings once the official announcement is made. Why? Because at that point, capable threat actors already have enough information to start developing working exploits, with or without our disclosure. Especially in the age of AI. Keeping our findings to ourselves doesn't stop them. It just leaves defenders with less information.

At the same time, we deliberately choose NOT to release our most advanced, weaponized exploits. We have exploits that work reliably in real-world conditions, even with unpredictable network traffic and other interference. Building that level of reliability still requires expertise that your average "prompt hacker" can't simply replicate overnight.

Releasing those exploits would meaningfully lower the barrier to exploitation and potentially cause real damage. We're not going to do that just to prove a point, get more attention, or win some internet arguments.

These aren't arbitrary decisions. They're deliberate tradeoffs between transparency, helping defenders, and not handing attackers capabilities they otherwise wouldn't have.

I know some people will disagree. And that's fine. We can't satisfy everyone, but I genuinely believe we're making the right calls.

We want to build a wildly successful company. We want recognition. We want to win. And we want to do all of that while making the security ecosystem better, not worse.

Those things aren't mutually exclusive.

One last thing.

Wake up. We're not living in the same world anymore.

AI has fundamentally changed the speed and economics of vulnerability discovery and exploitation. The assumptions we've relied on for years are becoming obsolete. The old timelines don't work anymore.

We all need to adapt.

Just like any real hacker would.
View quoted post on X
0 replies 2 reposts 11 likes 931 views View on X
Xint Verified account @xint_official
100% agree - from our recent falling cost of a hack research paper:
"A mediocre hacker can now use frontier models effectively and can expect the price to find such vulnerabilities to drop by 80% or more every 3-6 months. In other words, while we are still far from complete amateurs using unharnessed models to hack systems and applications, what used to be considered too hard and costly to be worth the effort of the average hacker is now reachable...

Expertise still matters.

None of the first runs on Opus 4.6, 4.8, and 5, and on GPT 5.6 Sol and GPT 6 Sol found the target bug unguided, so we had to rerun with more guidance measuring the original finding session as a reference cost. Software breaches still require some level of technical proficiency but no longer top-of-the-distribution expertise."

Full research report here (ungated, no need to provide an email) cdn.prod.website-files.com/69a605ca09bce7…
Marcus Hutchins Verified account @MalwareTechBlog 1d
If you hear anyone claiming "AI is enabling unskilled criminals/hackers to operate with the sophistication of elite state-sponsored hackers", you can safely unfollow them and ignore all their future takes on cybersecurity.

As someone who monitors threat actors professionally, and has watches thousands of unskilled hackers try to figure out how to hack with AI, it's basically just this:

Sometimes they succeed, a lot of the time they don't. But even when they do, it doesn't come within 1000 miles of anything resembling a sophisticated state-sponsored intrusion.
View quoted post on X
0 replies 2 reposts 12 likes 1.6K views View on X
Xint Verified account @xint_official
"Just patch faster" is not the answer. Patches are code too, which could introduce new own vulnerabilities or unintended behaviors. We recently hosted a webinar outlining a process that compresses timelines by taking advantage of what AI does well while minimizing risk to the overall system
xint.io/resources/shou…
xint.io AI-Assisted AppSec: How to Validate And Patch Vulnerabilities Faster Without Losing Control | Xint LLMs have made flagging possible weaknesses cheap, but proving which ones are real, exploitable vulnerabilities is not. Security teams are testing different automation strategies to accelerate...
Jeff Stein Verified account @jstein_sun 2d
NEWS: Wall Street execs and NSA have told Congress that outages were caused by patching vulnerabilities identified by AI models, sources say

AI did not cause outages but found so many issues that the rush to fix them before an attack overwhelmed systems

washingtonsun.com/technology/ban…
View quoted post on X