It makes to hold out for disclosure if there's a pay day involved. Since this is Google, the bug bounty is not the issue. I say release right away. Nothing else will compel the company being reported on to fix otherwise.
Perhaps eventually, they can finally live up to their namesake. On January 1 2029, we will move to a "Full 0" policy where vulnerabilities are disclosed after (looks at camera) zero days.
No, that's not what it means. It just means that the benefit must outweigh the risk, as always. Making surgical patches to existing code is also quite different to introducing features, which is what developers typically spend their time on. Not that same-day patching would imply literally zero time to review either.
That does not follow. Somebody stills needs to set it up. And maybe to sign off. The AI could instantly diagnose and summarize the problem and its remediation and you say the word.
No, they're not - they can just release and wait for another vulnerability disclosure. Then the loop between [Vendors] and Project Zero needs no humans in it.
If AI is doing all the work then why would remediation be needed at all? Why can't the software AI produces be vuln-free if the same AI is used to detect them?
You talk as though there is one AI model that is perfect and capable of identifying all potential flaws with no mistakes or omissions, and that access is free and unlimited.
Exactly. That's super irresponsible to knowingly release software that you haven't used AI to check if it had any vulns. There's zero (heh) excuse to release vulnerable software at that point.
Catching lots of bugs in software written with poor methodology is the new norm and that is not the same as releasing software that is not vulnerable. That is releasing software that needs increasing technological improvements or luck to
find remaining vulnerabilities.
I’m not familiar with this specific effort but it seems like the companies digging around for the bugs should also offer similar access to their products that discovered the vulnerabilities to the vendors or OSS projects to patch them. Otherwise vendors might use this as a leveraged approach to get people to adopt LLMs, specifically theirs, so their products remain reasonably attractive. Google, OpenAI, Anthropic etc could just turn on existing software as a strategy to get more adoption of their tools.
“Don’t like our constant stream of patches? For $10k/mo + tokens you can get access to the model to patch the vulnerabilities we disclose!”
For sure I would love to get access to P0 researchers for a m 10k/months, but i don’t believe you can hire such talent for this sum.
They are one of or maybe the most skilled security research team that’s doing public research and their blog is the source of many modern exploitation techniques
For a while, asking your agent to act like a P0 researcher was a nice nudge to find better bugs !
How is it irrelevant? Just because an effort existed before some change, doesn’t mean it can be repurposed and leveraged for other purposes. It’s the entire reason law and policy for government isn’t set in stone and is adaptable, because culture, technology and so on change.
The dynamic in this context absolutely has changed. Marginal cost to discover vulnerabilities from AI frontier labs and the big players has dropped significantly. The cost to patch is low, if you have access to certain resources. If you don’t, tough luck.
So this argument is essentially against the entire argument of AI alignment guardrails right now. Maybe you disagree with those, but I for one am happy they’re doing some things to protect against bio related warfare engineering. This is a case of cyber that’s about grey. Even there the labs have limited access to some and are choosing winners.
Usually you can’t find the kind of bugs P0 finds by just throwing AI at it. You need highly skilled people driving/guiding it. Highly skilled people are the bottleneck.
In any case it’d be sensible to make yourself familiar with the “specific effort” before you make up your opinion on it.
I mean, sure, the cited $10k / mo is exploitative for small open-source projects but... in general, why wouldn't we want there to be pressure on companies and projects to use high-quality LLM-based checks if they want to see wide adoption?
Ubuntu will need a LOT more time than that.
After seeing the latest vuln response times by Ubuntu, I switched my main PC to Debian; and will progressively migrate my homelab as I have time.
Definitely much happier. There’s nothing I miss from Ubuntu.
It makes to hold out for disclosure if there's a pay day involved. Since this is Google, the bug bounty is not the issue. I say release right away. Nothing else will compel the company being reported on to fix otherwise.
Starting today, our policy is “Full 90”, i.e. vulnerabilities will be disclosed 90 days after they are reported to the vendor.
On February 2, 2027, we will move to a “Full 60” policy where vulnerabilities are disclosed after 60 days.
On January 3, 2028, we will move to a “Full 30” policy where vulnerabilities are disclosed after 30 days.
Perhaps eventually, they can finally live up to their namesake. On January 1 2029, we will move to a "Full 0" policy where vulnerabilities are disclosed after (looks at camera) zero days.
If AI is doing all the work I could see that; same day remediation.
AI is not doing all the work.
Which means developers must trust AI blindly since there is no time to review the fix. Which also means that developers are completely unneeded
No, that's not what it means. It just means that the benefit must outweigh the risk, as always. Making surgical patches to existing code is also quite different to introducing features, which is what developers typically spend their time on. Not that same-day patching would imply literally zero time to review either.
That does not follow. Somebody stills needs to set it up. And maybe to sign off. The AI could instantly diagnose and summarize the problem and its remediation and you say the word.
And someone who understands the whole thing needs time to validate that the remediation is indeed what it's supposed to be.
No, they're not - they can just release and wait for another vulnerability disclosure. Then the loop between [Vendors] and Project Zero needs no humans in it.
“Someone” being another AI agent.
It’s agents all the way down.
If AI is doing all the work then why would remediation be needed at all? Why can't the software AI produces be vuln-free if the same AI is used to detect them?
Because the people finding the vulns aren't the people running the services.
Why aren't they? The people running the services can use the AI to find the vulnerabilities before they release.
You talk as though there is one AI model that is perfect and capable of identifying all potential flaws with no mistakes or omissions, and that access is free and unlimited.
Reality is more complicated than that.
Exactly. That's super irresponsible to knowingly release software that you haven't used AI to check if it had any vulns. There's zero (heh) excuse to release vulnerable software at that point.
Catching lots of bugs in software written with poor methodology is the new norm and that is not the same as releasing software that is not vulnerable. That is releasing software that needs increasing technological improvements or luck to find remaining vulnerabilities.
Is the AI in the room with you right now?
The Machine is everywhere. Watching us with ten thousand eyes. Listening with a million ears.
Unrelated, but didn't realise .google was a TLD.
Actually many brands pay for their own TLD, though most go unused (publicly).
https://en.wikipedia.org/wiki/List_of_Internet_top-level_dom...
There's lots of them now. Companies can apply to have their own for a few hundred thousand dollars.
so are .ads, .android, .cal, .chrome, .dclk, .docs, .drive, .gbiz, .gle, .gmail, .goog, .google, .guge, .hangout, .map, .meet, .play, .prod, .search,, .youtube, and .グーグル, .谷歌.
They are all owned by Google, but registrations are not open to the public.
I’m not familiar with this specific effort but it seems like the companies digging around for the bugs should also offer similar access to their products that discovered the vulnerabilities to the vendors or OSS projects to patch them. Otherwise vendors might use this as a leveraged approach to get people to adopt LLMs, specifically theirs, so their products remain reasonably attractive. Google, OpenAI, Anthropic etc could just turn on existing software as a strategy to get more adoption of their tools.
“Don’t like our constant stream of patches? For $10k/mo + tokens you can get access to the model to patch the vulnerabilities we disclose!”
For sure I would love to get access to P0 researchers for a m 10k/months, but i don’t believe you can hire such talent for this sum.
They are one of or maybe the most skilled security research team that’s doing public research and their blog is the source of many modern exploitation techniques
For a while, asking your agent to act like a P0 researcher was a nice nudge to find better bugs !
The "product" that finds Project Zero bugs is the people who work for Project Zero.
Project Zero has been around far longer than GenAI or LLMs, so your point is not really relevant here.
How is it irrelevant? Just because an effort existed before some change, doesn’t mean it can be repurposed and leveraged for other purposes. It’s the entire reason law and policy for government isn’t set in stone and is adaptable, because culture, technology and so on change.
The dynamic in this context absolutely has changed. Marginal cost to discover vulnerabilities from AI frontier labs and the big players has dropped significantly. The cost to patch is low, if you have access to certain resources. If you don’t, tough luck.
So this argument is essentially against the entire argument of AI alignment guardrails right now. Maybe you disagree with those, but I for one am happy they’re doing some things to protect against bio related warfare engineering. This is a case of cyber that’s about grey. Even there the labs have limited access to some and are choosing winners.
Usually you can’t find the kind of bugs P0 finds by just throwing AI at it. You need highly skilled people driving/guiding it. Highly skilled people are the bottleneck.
In any case it’d be sensible to make yourself familiar with the “specific effort” before you make up your opinion on it.
You say that like it's a bad thing...
I mean, sure, the cited $10k / mo is exploitative for small open-source projects but... in general, why wouldn't we want there to be pressure on companies and projects to use high-quality LLM-based checks if they want to see wide adoption?