Welcome to the Strip Mining Era of OSS Security(metabase.com) |
Welcome to the Strip Mining Era of OSS Security(metabase.com) |
What is true is already so.
Owning up to it doesn't make it worse.
Not being open about it doesn't make it go away.
And because it's true, it is what is there to be interacted with.
Anything untrue isn't there to be lived.
People can stand what is true,
for they are already enduring it.
Imagine somebody finding a flaw in a mathematical proof and everybody being sad because a beautiful proof got invalidated rather than being glad future work won't build on flawed assumptions.
I get that the rate of vulnerability discovery can be a burden, especially for people doing FOSS in their spare time, but the sustainability problem with that has always existed and only gets exacerbated by the vulnerability stuff, but the latter isn't the cause you need to make go away.
- A bug exists and nobody knows
- A bug exists and some people know
- A bug exists and everyone knows
As an outside observer, there is no way for you to determine if a bug is in state one or two, you only know once it's in the third state.
Which is the entire problem here. Having the bug be known to everyone is a vastly improved state over being known to a few. Yes, the bug being completely unknown is better than being known to a few, but there is no way to ever know if that's the case.
From the outside, known to none and known to a few are indistinguishable, and thus both states are the worst possible case. The only remedy is to make the bug known to everyone such that it cannot be covertly exploited.
Is your assertion that, since you specifically didn't know about the bugs that nobody, not in Russia or anywhere else did?
Obviously if bugs are out there existing in software and you don't know about them, or the CVE system doesn't know about them, or whatever ... this does not preclude bad guys from knowing about them. In the era of agents, knowing the bug exists is equivalent to having a PoC, so the distinction completely collapses.
Sweeping things under the rug is how we get insecurity. Sunshine is the best disinfectant.
Is this supposed to be hard to imagine? I can completely imagine this, especially if the mathematician is a celebrity in their field.
In practice, how much effort it is to find vulnerabilities matters a lot. We're in a time where things that used to be quite hard are now easy and the rate of discovery will change.
This rate of discovery matters a lot -- for OSS maintainer burnout if nothing else.
I don't think anyone is saying that here.
I think the net result is "wow, we're going to end up a lot more secure in several months, but things are going to feel sucky because stuff just got A) way easier for the average bad guy, and B) way busier on the fixing side."
I think it's likely we end up with an equilibrium with a lower rate of bug discovery than we're used to, but we need to experience an above average rate for a long while first...
The philosophy in this subthread may be too deep for me.
Me and the Jedi at the ends of the bell curve are just thinking "It's bad when your attackers know your code is vulnerable"
open source will get fixed while closed will not.
The patching cycle can become a problem for certain operations / industries.
Everybody hates the work, and security is often seen as a barrier and a cost center, not a driver or revenue.
Try binge-watching old Star Trek episodes, to see how Spock deals with the illogical 99.9% of humanity?
It's gotten much easier to reverse engineer binaries in general, and security patches in particular. Basically, an LLM can turn binaries into 'readable' code, and then reason about said code.
But yeah, if you're distributing binaries publicly, then you're going to have very similar problems.
This understanding may be incomplete or outdated (things moving very fast right now). I'd love to hear from a someone with more experience using LLMs to do binary analysis about the level of 'binary annotation' needed for LLMs relative to humans.
One of the benefits of Open source has been that there are more eye balls on the source, leading to more secure code/better quality. I think given enough time the bug reports will plateau and we will be back to a normal cadence - once the tsunami is over, hopefully things will settle at a more manageable cadence .
OSS has always had tradeoffs and I sadly think this one is going straight to the "Cons" column. We still think the Pros outweigh the Cons, but this is NotGreat.
Source that is unmaintained is dead. Nobody is looking at it, even the maintainer has something better to do.
Do you know whats even more powerful than "eyeballs"? Money.
Won't matter if is closed source, signed, and or obfuscated. =3
* I presume I'm not the only one to find the agents tasked with adding unit tests will sometimes try to sneak through "open source code and apply regex to confirm presence or absence of specific string literal".
They can speed you up significantly, but you absolutely do need to pay attention to what they produce.
I'm sure what they have is awesome, but it's clear that there are people out there with some decent prompts that are getting results out of widely available models as well.
The big thing we're sharing is: bulk scanning by random people in random geographies got a _lot_ better around January, it's widely distributed, and it's going to get a lot better regardless of whether that specific version of Mythos becomes widely available or not.
Besides that, hiring a beefy GPU instance at Vast.ai or similar places then running your own uncensored models on it, I've had great success with AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-NVFP4, smart + uncensored, but there are lots of options, probably some are already tailored for security research.
I take it that Metabase is both not paying bug bounties and not using these tools internally?
If that's the case, Metabase is not going to get meaningful investment from researchers who want to fix issues, but they'll get increased attention from malicious attackers who have no qualms exploiting the vulnerabilities for profit.
LLMs have made it a lot easier for people to find vulnerabilities in software. Open-source makes it easier, but we already have non-AI tooling (IDA Pro, Ghidra) that's good at binary reverse engineering, and LLMs can use that output to find vulnerabilities as well.
This year, as I select products to use for sensitive data, I've been paying a lot more attention to whether they offer bug bounties and for how much. For example, I like Kagi for search and thought about trying Orion, their web browser. Then, I saw that Kagi's been paying $100 for UXSS vulnerabilities.[0] For comparison, Firefox pays $8-10k,[1] and Chrome pays up to $10k for the same class of bug.[2]
[0] https://help.kagi.com/kagi/privacy/bug-bounty-program.html
[1] https://www.mozilla.org/en-US/security/client-bug-bounty/
[2] https://bughunters.google.com/about/rules/chrome-friends/chr...
The interesting thing is that the business model seems to have changed. Why collect a 10k bounty when you can advertise a 3k/month scanner?
At the risk of repeating myself -- this is targeted at other OSS maintainers, not random people who might have done a git pull of some random project a couple years ago.
I do think security is going to require more, not, less human investment as attackers may be running automated vulnerability screens from the outside that you must counter, as well. Without rigorous internal processes to manage and screen all changes and upgrades, companies risk leaving themselves open.
One design change which limits exposure is to have more local-first apps or experiences so there's less cloud / server to computer interactions to secure.
Umm... no? It's called OPEN source. Expecting people to cancel their plans to make your free software more secure is pretty audacious. Luckily, many WILL, but the expectation is just foolish.
These alerts are absolutely not being shared publicly before we have a fix for them.
MIT:
THE SOFTWARE IS PROVIDED “AS IS”, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
BSD:
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS “AS IS” AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
OSS maintainers don't owe anyone shit. Anyone who thinks fixing the bug is important is free to fix it and submit a patch.
Or you know, provide the security companies and businesses using your software for free with all the fix timelines and out of hours support they’ve paid for (none).
In theory, nothing.
In practice, it's in our long term interest that bad things don't happen to them.
How sustainable all of this is, I have my doubts?
I have dog-fooded it heavily on my own projects, client projects and friends projects. It finds things that are really quite clever and not obvious. It really helps me.
But when I try to do the obvious thing for sales of using an OSS project to get hype, show off etc. I find that it becomes really hard to really know that I am helping and not just spamming.
To be clear - I think for an AI tool like mine to actually give you clever results that finds not obvious issues and security flaws - it needs to have some level of false positives.
I find myself struggling to justify the approach of firing off defects to an OSS maintainer without verifying them - which takes considerable time if I am going to do a good job. Even with tools to help pull apart the code, the core problem is always you don't know what you don't know.
The same process working on my own projects I can eat through a ton of defects and find some really great stuff. But that's only possible because I can tell at a glance what is real, what is fake, and also what is an oh ** issue.
So I think this is true, but the risk is that people who don't understand the projects just point scanners at OSS blindly and ruin the good work maintainers are doing.
This stuff is more complicated than people give credit - and it's so easy to kid yourself into thinking any bug report is helpful.
And you're surprised OSS projects are pivoting towards "open source does not mean open contributions"?
How do you get that from:
> the risk is that people who don't understand the projects just point scanners at OSS blindly and ruin the good work maintainers are doing... and it's so easy to kid yourself into thinking any bug report is helpful.
Defining an "era" as a "summer" is short-sighted. Calling an industry-wide efforts to find and fix security vulnerabilities with better tools "strip mining" is backwards thinking, from where I sit.
People who prefer 0days in their code baffle me.
Ignore (admittedly low-effort LLM generated) reports at your own peril.
Fact is that Mythos found only one issue in curl and nothing at all in most code bases. It is getting quiet around Mythos, and the AI companies will move on to the next scam.
In most open source projects, Mythos or similar tools have found nothing. The AI people only contact the projects where they find something, because it would be bad for marketing otherwise.
As a reminder; your account has been shadow-banned, it looks like you got a little unlucky in 2016.
Absolutely, and the "false-positive" issue people keep citing as why Mythos is so good is easily solved in the harness, simplest solution is starting fresh context with another prompt to evaluate if it's a false-positive or not, just adding that drastically cuts down the rate.
Fun times =3
That makes sense. Java leaves metadata in compiled code to reconstruct better.
> your account has been shadow-banned, it looks like you got a little unlucky in 2016
I see. I wasn't aware. Thanks. I thought the two comments per day limit is for users with low karma. Anyway, I'm using that limit to push myself in being picky about posts to comment on and to improve the quality of comments.
I'd recommend creating a new account. Or you could email the mods (I forget the email but it is publicly available somewhere) if the user name has sentimental value. In your case it is clear that a shadowban is inappropriate.
If software "is a trap", even my ever-computing loving wrote first programs on an Apple II in the 80s will only be as you sort of describe invested in by reference (minimal usage).
But no-one will sign up for a "trap" as a career, and only those who do will deal with its problems. The first thing that comes to mind is "Johns", "Hotels", and the trappings of the sex trade.
Whether a bug is exploitable is an entirely separate category of unknowable, because seemingly-innocuous bugs quite often have very deep and very subtle implications that when combined with another innocuous bug, result in an RCE or PE.
Therefore, it's sensible to treat all bugs as potential threat vectors unless and until proven otherwise. Which brings us full circle: state 3, all bugs being public, is probably the safest thing because nobody can know if a bug is in state 1 or 2.
Sure, who wouldn't like to have that? Such a thing is impossible to reach starting with the same reason as Gödel's incompleteness theorem is a thing, plus a gazillion of more practical constraints.
Who gave them "the right to scan"? You did by hosting your open source in public. But scanning a public service prior to AI was still covered by "Unauthorized System Access".
But what if they are wrong, and given the self-serving nature of these scans, now your repo is just OJ Simpson? And your software is banned due to an external scan you did not ask for?
Is there no one in this world who will be accountable for any thing at all? Can we sue the scanners if they are wrong and publish their results for defamation even in a public PR?
These things will happen. IF I had source in the open and a scan result was incorrect that nobody asked for and the results had false positives, I would sue Anthropic for defamation and I would win.
With careful prompting, LLMs will give up some of their sources and methods. Claude describes the legally and ethically suspect methods Anthropic used acquiring training materials for its models.
The IP law and courts are starting to catch up (re: Anthropic settlement September 2025), but licensing language and enforcement has not.
You're also assuming that they haven't made the alternative judgement that instead of triaging the haystack of slop that they get in order to potentially pay out to someone, they should instead be spending that cash and effort on tokens to find bugs in their own codebase.
The claim I'm rebutting is "in the past few months nearly every LLM generated report is real." If that were true, there would be no need to close the bounty. The bounty is to address approaches that they themselves may not have considered, so would still hold value if the claim held true, as outside individuals may still hold unique LLM-assisted approaches and perspectives.
Burnout means that no more fixes come - ever - and that things sit vulnerable until everyone relying on that tool takes the time to build and switch to a replacement.
Maintainer burnout is perhaps the single biggest threat to the ecosystem right now.
Pay each maintainer an absolute minimum of $200K a year or shut up and do the work yourself - in a fork if necessary.
And if the capacity is overshot (which I believe is happening as we speak), users end up in extended states of being insecure.
I'm also one of the unwashed rabble who believes there is a large practical difference between a vulnerability that exists but isn't found and one that is widely known and exploitable.
If you believe people don't use software that is unmaintained and hilariously out of date I genuinely don't know what world you live in or how to deliver the bad news to you.
There is no right to demand someone does something for free, and we have gotten dependent on people doing things for free. We don't have to pay people but if we don't want to, then we have to be willing to do it ourselves. Otherwise it could go away at any moment and we have no recourse.