Hacker News

I think a tricky thing about these metrics is GitHub and the Linux Kernel became their own CNA in mid-2024[1] and mid-2025[2]. This caused a bunch of new automation and lowered barriers to actually filing a CVE.

As a small data point I have a >20k downloads Obsidian plugin and it got a CVE on a Beta release due to a bad pathing regex. I feel this would have never happened 2 years ago.

So, I feel the bar for a CVE being issued is WAYYYY lower in 2025+ than before.

[1]: http://www.kroah.com/log/blog/2024/02/13/linux-is-a-cna/

[2]: https://github.blog/security/vulnerability-research/how-to-r...

I was suspicious the absence of a severity filter was exaggerating the trend...

But then I found this:

https://craci.com/resources/CVE-publication-trends?sev=criti...

Trend for critical CVEs is actually worse, at roughly 25x that of 2020, compared to all CVEs which are roughly 5x compared to 2020.

It’s impressive that the number of CVEs issued the last two years has skyrocketed, given how badly DOGE gutted the cybersecurity [0] agencies. Unfortunately we don’t know how huge the backlog may have since grown

[0] https://www.securityweek.com/mitre-signals-potential-cve-pro...

Cool. This would be easier to interpret with a log scale. There's an option for it, but clicking it doesn't change the chart and freezes the site for me.

Others have pointed out that we thought we would get security vulnerability from AI code, but it turns out AI is uncovering vulnerabilities in decades of human code.

Not the topic (the site is interesting), but a tangential rant.

I think the whole CVE/CVSS system is completely outdated, and it doesn't reflect the reality of software making for a while now (years). Also. for years, major software providers have been fighting to control their own release logs and security vulnerabilities. Recently, NIST stopped actually analyzing CVE's (one can relate to that), and the fact is that the system only makes sense as 1) vuln database; 2) in the context of responsible discosure; Responsible disclosure is from a non-AI era, where the time between a finding and a release woulds be measured in months, not weeks. Also, from a time where the probability of another human to stumble on that specific vulnerability was quite low. Pre-AI times.

Also, in April 2025 MITRE announced that fundings to the CVE/CWE program would expire, and it took a couple of days for the Trump administration to backpedal from that. We need a global system, not an american one - and clearly, americans should not be in charge of these global programs, as there is no way to isolate the world from their local politics. Their programs, their funding, nothing against it. I just think we, collectively, should have something a bit more robust, and while at it, do it on a quite more modern basis (EU CRA is actually a good start, as it includes SBOM and mandatory reporting).

CVE, CVSS, CPE seem all designed by comittee; CVSS and CPE are a flagrant case of that (try to apply CPE versioning to your own software, you'll get it); Its "what we have now" - there are alternatives, afaik but not that mainstream.

I remember a comment somewhere, where employees of a company, would open a CVE for a piece that they use in their job, to force an upgrade of said software since normal update request has been denied despite having business reasons to do so. So, the number doesn't matter.

Also, the linux kernel assigns CVE for anything that can be though as a vulnerability:

> Note, due to the layer at which the Linux kernel is in a system, almost any bug might be exploitable to compromise the security of the kernel, but the possibility of exploitation is often not evident when the bug is fixed. Because of this, the CVE assignment team is overly cautious and assign CVE numbers to any bugfix that they identify. This explains the seemingly large number of CVEs that are issued by the Linux kernel team

In other words, the number is fake

I prefer the tornado

https://rogolabs.net/tornado/

Thanks for this. I’m going to send it on our #general on Slack so people might understand when I say I’m a “bit” busy at the moment…

I am more interested in the long term trend. Do we burn through the low hanging fruit at some point or is this the new normal?

The Shigure mode cracked me up

You'll sleep easier if you think of the graph label as "vulnerability removals"

12,000 CVEs? That doesn't match the data in the source at the bottom. I cloned it and for 2026 there are 79,764 CVE json files. I'm pretty sure I patched more than 12,000 CVEs for my RHEL systems this year. Also the site freezes for me as well when switching to logarithmic.

CVE moonshot and no one is getter paid. How very 2026.

This is what AI companies meant by embracing the exponential?

Are folks finding it relatively easier to get SRE positions now? Demand ought to be up.

Avalanche on Mount Tech Debt?

Looks like an exponential curve, the inflection point doesn’t seem to line up with either the introduction of AI, or AI’s inflection point. Yes, yes, lower your prompt box, Software Development Has Changed Forever In The AI Era, but this is just “CVEs got Goodharted”.

It has hit the point where I’ve told people reporting bugs with my software that I won’t acknowledge the bug unless they provide a patch to mitigate the issue.

A lot of these bug reports are incorrectly filed. Let’s look at one issue, which I am sure was found with AI assistance:

CVE-2026-40719

Here’s the issue: An IP authorized to perform recursive queries can cause the DNS server to be non-responsive if it sends hundreds of specially crafted packets a minute (this can be changed to millions of packets a minute if one sets a parameter in the configuration file). Normal service is restored within a minute of the attack being stopped. The workaround would require me to essentially rewrite the recursive part of the DNS server.

Since the DNS server, like any server in a LAN, is already vulnerable to a flooding attack, this is just a flooding attack with extra steps.

I told the guy that was interesting, I’m not being paid enough to rewrite the DNS server to be immune to the attack, and people shouldn’t be using its recursion (it has very limited IPv6 support, for starters), and if they do use its recursion, only authorized IPs (e.g. 127.0.0.1 or 192.168.1.0/24) should be allowed to perform recursion.

It wasn’t a serious enough attack to merit a patch, and anyway the guy reporting the bug didn’t even have the decent courtesy to give me a patch.

The guy responded by filing for a CVE report, incorrectly calling it a “high impact” bug, and I had to make a “patch” which simply disables recursion: https://samboy.github.io/MaraDNS/patches/MaraDNS-CVE-2026-40...

Point being, a lot of these CVE reports out there which supposedly are high impact bugs are trivial things like “someone can overload a server for about one minute if they send hundreds to thousands of specially crafted DOS packets”.

(Today, I got an issue which, as far as I can tell, looks to be a Linux kernel issue and not an issue with my code. I told him to show me how he would patch my code to “fix” the issue, but so far have gotten nothing but crickets)

In case you didn't notice this is yet another LLM slop project, have a look at the sourcecode, it's your typical everything-in-one-file regurgitation with loads of very verbose, claudish comments...

what would help with this is to regulate and restrict frontier models from doing cyber work so that the good guys are defenceless against bad actors /s.

really disagree with western policy makers and ai company leaders on the path they are steering us, leaving us defenceless all for the sake of their obvious self serving attempts at regulatory capture.