The AI bugpocalypse is here. Now what? - Jack Cable, Corridor

AI Engineer · 19 min · 169 sentences · from YouTube's caption track

Each timecode opens YouTube at the start of that sentence. Line anchors (#s42) are the cue ids in the WebVTT, and every line carries its start and end seconds. All transcripts has every talk, and the whole corpus as one file.

00:01Hey there, I'm Jack Cable and today I'm going to be talking about the effects of the AI bug apocalypse.
00:06As you may have seen, frontier models are getting better than ever before at discovering and exploiting vulnerabilities in our software.
00:14Right, this is leading to what many are calling a bug apocalypse where we're finding more and more vulnerabilities, particularly in the open-source libraries that power um all of the software we rely upon.
00:26Right, so today I want to break down what exactly is happening and how defenders can get ahead of the exploitation that is occurring.
00:35As far as my background, right now I'm the co-founder and CEO at Corridor, a company I started about 18 months ago focused on securing AI coding.
00:45Before this, I served as a senior technical advisor in government at CISA, the Cybersecurity and Infrastructure Security Agency, where I worked with top software companies to help them build their products to be more secure by design.
00:59I'm also an ethical hacker.
01:01I got into the top 100 rank of hackers on when I was in high school.
01:05I studied computer science at Stanford.
01:07So I've seen firsthand how these simple, you know, repeat classes of vulnerabilities can be introduced and exploited and have been um close participant right in many of the the most recent
01:20um advancements and seeing just what this means uh for both our adversaries as well as defenders.
01:28Just to set the stage, right, as everyone here knows, I imagine AI coding tools are scaling faster than any software category in history.
01:36Cursor, Copilot Code um grow exponentially and with that, right, also comes these improvements in how frontier models can find and exploit vulnerabilities.
01:48Um so we're seeing, right, both ends of the equation shifting.
01:51On one hand, um models can do better job finding vulnerabilities.
01:55On the other hand, our attack surfaces are growing immensely, right, as AI becomes the default code writer.
02:03Um so, what I want to explore in this talk, right, is how do we balance that?
02:06How do we make sure that we're not going to have immensely more vulnerabilities than we've ever had before, right?
02:12And just to to give some sense, right, I'll move myself here of um some of the statistics, right, um pulled some from last year where about 84% of developers were using AI coding tools, 30 to 40%
02:25of companies encouraging use of AI coding assistants.
02:28That was from Stack Overflow, right?
02:30I don't I haven't seen the latest numbers this year, but what I would expect once those come out, right, is that is the vast, vast majority of developers and companies
02:39who are using coding agents, right?
02:41And part of this is the increasing level of autonomy by which these coding agents are being used.
02:48It's no longer, you know, auto complete, often it's not even a developer synchronously within cursor.
02:54Um right, when we do our own development, right, now it's um spinning up agents from within Slack or wherever folks are working and having many agents run at once in the background.
03:06Um this is a tremendous shift in how software is being built, um and at the same time, right, like I mentioned, the frontier models are getting significantly better.
03:15Um and you can look at it from, you know, pretty much any part of the cyber attack chain, ranging from finding vulnerabilities where models can, you know, now do do,
03:26you know, significantly better than even I could, and you know, I've I've reported hundreds of vulnerabilities to various companies.
03:32Um so, everything from finding vulnerabilities to exploiting them.
03:36Uh this is a chart here that comes from Anthropic, right, showing Mythos compared to a number of um other models that they and others have put out.
03:45Um and we can see that we're we're seeing quite rapid advancements in models' capabilities, right, And and particularly to execute more kind of autonomous attack chains.
03:58So, as we think about adversaries who are using these models, right?
04:02They're not just going to be discovering vulnerabilities, but they're going to be automating every part of the attack process.
04:09So, it's our job, right?
04:10As defenders to understand, okay, what are the points where we can make software systems more resilient to all of these attacks.
04:19Right?
04:19And to me, this brings back a lot of the work that I was doing in government around the the secure by design initiative, right?
04:26Um and so so the overall question, right?
04:28That I'm worried about is how can we make sure that frontier AI models aren't introducing exponentially more vulnerabilities over time, right?
04:37Um even pre-AI, we've had this you know, uh heavy increase in common, relatively simple classes of vulnerabilities that are being exploited by adversaries.
04:49AI is making this significantly easier, right?
04:51So, I think the only way that we're going to to win as defenders is if we use the the same techniques, right?
04:58To harden our systems.
04:59Um and I would say that there is good news here, right?
05:01That a lot of the vulnerabilities, pretty much all of the vulnerabilities that even frontier AI models are finding aren't anything new.
05:10Yes, it's new that uh given vulnerability was found in a you know, specific file with with a piece of software, but that vulnerability class um isn't necessarily novel.
05:21Um and and we can actually use that to our advantage, and I'll I'll get into that.
05:25Um right?
05:26So so overall, um the the thesis here is that um right?
05:31We are seeing both attackers get more tools in their toolkit, and the same time, the way in which software is being built is fundamentally changing.
05:40Um so so really the question then becomes how can we apply AI to shore up on software systems?
05:47Um right, and I want to to take a quick detour um to some of the secure by design work that that I I kicked off with with others in government.
05:57Um right, this is a paper that we put out in March of 2023.
06:02So, just as you know, LLMs were starting to become more readily available, but um right, their application in coding at that time wasn't much more than, you know, auto complete.
06:13And while that's useful, it wasn't necessarily the step change that we have now.
06:18Um and what we focused on kind of laying out with this vision, right, was this idea that it it isn't really rocket science when it comes to preventing vulnerabilities
06:28in software, right?
06:29While it's true that it's hard to build a perfectly secure system, we do know how to build systems that are fundamentally more resilient to common classes of vulnerabilities.
06:40Um and and just to, you know, make this concrete, um this is a set of vulnerability classes coming from MITRE.
06:48It's the uh top classes that are um exploited in CISA's known exploited vulnerabilities catalog, right?
06:56And if you go down this list, you'll you might notice, right, that pretty much all of these are basic types of vulnerabilities that not only have we known about for decades, but we've known how to prevent at scale for decades,
07:09right?
07:10Uh take buffer overflows, right, number two on that list.
07:13Um and by the way, these are the same vulnerabilities that models like Mythos are finding in software.
07:19Um and buffer overflows were first documented um about um I believe 30-plus years ago, right?
07:27Um so we've had uh documented instances of how to find and exploit these vulnerabilities, and we also now have languages that are memory safe, right?
07:37Languages like Rust, Go, pretty much any language um that that's not C or C++ um is built in a way such that it's impossible to introduce memory safety vulnerabilities, right?
07:50They have guarantees that prevent those from being introduced.
07:54So we have techniques by which we can prevent them and yet they continue getting introduced over and over again, right?
08:03So so let's look at memory safety for instance.
08:06There's you know the statistics range but approximately 60 to 70% of vulnerabilities in products written in memory unsafe languages can be completely prevented using memory safe languages, right?
08:20And this is you know based on CV data out there and not only that right we've seen a lot of companies Google, Microsoft, Amazon even you know open source software
08:31the Linux kernel is being rewritten in parts in rust.
08:36We've seen real evidence that by shifting to memory safe languages you can reduce overall vulnerabilities, right?
08:43On the right here is a chart from Google showing the rate of memory safety vulnerabilities over time in the Android operating system.
08:52What's interesting right is that this isn't even you know they're they're not even necessarily rewriting code in a memory safe language.
09:00They're just writing new code in a memory safe language and even then right the percent of memory safety vulnerabilities has dropped quite dramatically from you know about 75% in 2019
09:11to maybe 30% in 2022.
09:15So to me that's personally quite exciting, right?
09:17Because it means that's not a given that we're going to continue having these basic vulnerabilities over and over again, right?
09:24And in part of the you know high-level policy conversation I think as a result has to be not just how can we deploy these frontier models to find one-off vulnerabilities in software
09:35that is something that we should be doing but at the same time, right, I don't want to miss out on opportunities to make our software fundamentally more secure.
09:43Right, we could pour millions of dollars into essentially playing whack-a-mole with vulnerabilities and patching them one-off in some of the open-source libraries that we all rely on, or we could do a one-time rewrite, for instance, to move some of these critical libraries
10:00into a language like Rust, right?
10:03And then that will pay dividends for years to come.
10:07So this is really at the core of how I'm thinking about this, right?
10:10Is what are some of the fundamental changes that companies, that open-source developers can be making that can reduce exploitation both by models today, right, and models to come.
10:22Cuz the advantage, right, of doing a rewrite, for instance, is that if you have some of these fundamental guarantees, then even if the models get smarter, right, Rust has programmatic
10:33guarantees such that we know that memory-safe D vulnerabilities in no circumstances won't be possible to be introduced or discovered.
10:42Right, and all of this comes in the context, too, of the fact that AI is increasingly capable at, of course, both writing code and then introducing vulnerabilities.
10:53You might have seen a couple months ago one example where Opus 46, right, by all accounts a very smart model, introduced a vulnerability in a smart contract that led to a couple million dollars being stolen,
11:06right?
11:07So so while the models are very smart and capable, often times security is very contextual.
11:13And the model just might not have the context in order to know that it's introducing a vulnerability, right?
11:19And this is reflected in academic benchmarks.
11:23One, for instance, here, Backsbench, you can find that backsbench.com by researchers at ETH Zurich, UC Berkeley, uh finds that even the best models introduce vulnerabilities about 20 to 40% of the time when writing code.
11:39Right?
11:40Um and this shouldn't necessarily come as a surprise.
11:42Um for one, right?
11:43Models are trained on all of the world's existing code, and humans haven't been great at not introducing vulnerabilities in code in the past.
11:51But two, right?
11:52Increasingly, and this kind of lines up with some of what we're seeing among our customers, is that the vulnerabilities being introduced are often and less so the basic one-liner
12:02vulnerabilities, and more so contextual issues.
12:05Right?
12:06Things like authorization bugs that require an in-depth understanding of a company's business logic.
12:12Um and that's something, right?
12:13Even if the model's very smart, it's not being trained on your company's proprietary information or how your own um kind of, you know, threat model works.
12:24Um and that is why, you know, I believe we're still seeing quite a high rate of vulnerability introduction um even by um, you know, by by all accounts very intelligent models.
12:36Um so so let's now think about, okay, given that the vast majority of software development is being done with AI, how can we make sure that AI is capable of writing secure-by-design
12:47software?
12:47Right?
12:48And part of this is a shift we're seeing in the level of autonomy um that AI is um now given when it comes to software development tasks.
12:58Right?
12:58We're kind of moving up this ladder that started with auto-complete to you know, agents within Cursor, Claude Code that can synchronously produce code, to now increasingly these autonomous agents that can work for an hour, hours at a time,
13:13and produce quite large code changes.
13:15Of course, the next step then becomes agents that are reviewing code.
13:20Um and we we at Quarter believe that, you know, within the next 6 to 12 months, the majority of code that is being shipped will be reviewed uh not by human but by AI.
13:31Um I think that's just a kind of natural consequence of the rate at which companies need to move, given that code review is not is now the bottleneck and I don't think we're going to accept that
13:41for very long.
13:44Um so really, right, our perspective at Corridor is around preventing vulnerabilities before the pull request, um as well as giving visibility into how AI coding tools are being used.
13:53Um and I think that's really essential, right, is that security cannot be the blocker when it comes to companies accelerating their development, right?
14:04Um acceleration is always going to win out.
14:07Um so when we talk to security teams, the conversation is less around should you allow your, you know, development teams access to coding agents?
14:16The answer is obviously yes, right?
14:18It's more around how can you do that with guardrails in place, right?
14:21Because what we're seeing is that without guardrails, yes, the coding agents can introduce vulnerabilities, um and in order to get to a point where, you know, development can be more autonomous,
14:33that code can start to be reviewed um by AI and merged in without as much human oversight, we really need to have tooling in place that allows security teams to have that assurance,
14:47um and to to give the blessing to their their engineering team to accelerate.
14:52Um I I want to close with some of the policy perspective, right?
14:57And um this is in part tied to the recent export controls on Mythos and Fable models, right?
15:05As part of a letter uh led by Mikayla Gyalcsamis, um where we urged the White House to lift the export controls on these models.
15:16And the perspective there is that the benefit to defenders far outweighs the risk, right?
15:25These are very powerful and let's face it, right, dual-use models that can both be used to secure systems and also um to exploit them.
15:35Uh to Anthropic's credit, they have done a lot of work with the Fable release, right, to have some safeguards in place such that um they're more skewed towards defenders
15:45than than adversaries.
15:47Um but this is also coming in the face of, right, increasingly powerful open-weight models.
15:52Um you've probably seen, you know, documented um distillation attacks um where open-weight model providers can train on the output of um closed-weight models and that um as a result is quickly shrinking
16:08the, you know, timeline between when a frontier closed-weight model comes out and when open-weight models catch up to that.
16:15Um so whether we like it or not, right, adversaries already have access to incredibly powerful models um and they're already using them today to exploit systems, right?
16:26So to me it becomes more question of how can we rapidly um get the kind of capabilities in the hands of defenders and I think that requires having these models be more widely available.
16:40Um one cool thing I had the opportunity of doing a couple weeks ago was testifying to uh the US Congress um on the, you know, risks of both uh frontier models
16:51as well as AI coding.
16:53Um my recommendations had a couple elements, right?
16:56One was to prevent vulnerabilities in new code going forward.
16:59I think this is something that both every company as well as the US government should be focusing on and making sure that as development accelerates, right, security isn't being left behind there.
17:10Um second is to harden the open-source foundation.
17:13Um this is incredibly important, right, especially since open-source software is going to be the kind of proving ground for a lot of adversaries, right, who want to test out these models and exploit vulnerabilities they find
17:26due to, right, the exact nature that's open-source.
17:29So, so you can just go and run a very smart model on them and in all likelihood find many novel vulnerabilities.
17:36Um so, so I think both the US government, but then private companies as well have a responsibility to help shore this up.
17:43And like I mentioned, right, it's not just about one-off, um you know, vulnerability discoveries or patches.
17:49It really has to be more systemic and start to get into um rewrites that can fundamentally reduce the risk of vulnerabilities that can be found whether by models today or in the future.
18:02Um and then the last area of my recommendations were to foster an ecosystem of American weight made open-weight models, right?
18:10I think in order for us to stay competitive here, it can't just be closed-weight models alone, right?
18:18Um there's a number of reasons here, you know, one of those is that for many companies, while there's a place for for closed-weight models, you also might want to do things like fine-tuning models.
18:29Um and that is only possible with an open-weight model.
18:33Um so, so I think it is really essential, right, that um we have frontier open-weight models coming out of the United States.
18:41We haven't seen as much of that to date, um but I think that's a critical element of American competitiveness when it comes to AI.
18:50Um so, so that those were my overall recommendations, right?
18:53And of course, all of this is in the context of these increasingly powerful models.
18:58The hearing was, I believe, a few days before, you know, um Mythos and Fable came out.
19:04Um and then, of course, all of the um export control um actions that were taken.
19:09Um so, this is an incredibly rapidly evolving space, but I think that's why it's all the more important to go back to the basics, right?
19:18What are the fundamental controls that can protect against um any vulnerabilities that can be discovered by models today or in the future?
19:26And I think that's where we really ought to be spending our time using these models um to make our systems more resilient.
19:35Um, so that's my talk.
19:37Um, happy to be reached.
19:38My email's here and thanks everyone for tuning in.