FBI Says Artex AI Open-Source Hacking Tool Used to Breach Networks In 'Under Three Hours' (techcrunch.com) 22
The FBI is warning (PDF) that hackers are using the AI-powered penetration-testing tool Artex to compromise networks in "under three hours" in some cases. The tool has been linked to attacks against South Korean financial institutions, at least 2,250 ConnectWise customers and Barracuda. TechCrunch reports: Artex is one of several open-source AI-powered penetration testing platforms, a category of software built to probe a network's defenses. Hackers can connect Artex to hosted frontier large language models (LLMs) like those from Anthropic and OpenAI, as well as to open-source models and non-U.S. models. The FBI says Artex intrusions can be "noisy," meaning they generate reams of traffic and data that may trigger a targeted company's defensive alerts. But the tools allow hackers to move at speeds that makes it far more harder for human defenders to keep up.
In some cases, Artex-linked attacks can move from targeting a system to active compromise "in under three hours." Oftentimes, Artex will rely on rapidly scanning a victim's network to look for signs of weakness, and then exploit simple security bugs, such as weak protections and exposed or stolen passwords to gain a foothold. From there, Artex allows the hacker to spread across the network for databases, mailboxes, and other sensitive information. [...] After Artex was used in cyberattacks, the tool's developer made the software closed-source to prevent further attacks. But the FBI says the previous open-source version continues to circulate and remains "widely accessible" for use in cyberattacks.
It's not only companies and targeted victims that have been compromised as a result of weakened security exploited by AI tools; the tools themselves have been exposed as well, potentially allowing outsiders to access powerful hacking capabilities. Last week, researchers found hundreds of publicly exposed Artex systems online, which had been placed on the internet without a password, likely by mistake or misconfiguration -- including the Artex system that was used to target South Korea.
In some cases, Artex-linked attacks can move from targeting a system to active compromise "in under three hours." Oftentimes, Artex will rely on rapidly scanning a victim's network to look for signs of weakness, and then exploit simple security bugs, such as weak protections and exposed or stolen passwords to gain a foothold. From there, Artex allows the hacker to spread across the network for databases, mailboxes, and other sensitive information. [...] After Artex was used in cyberattacks, the tool's developer made the software closed-source to prevent further attacks. But the FBI says the previous open-source version continues to circulate and remains "widely accessible" for use in cyberattacks.
It's not only companies and targeted victims that have been compromised as a result of weakened security exploited by AI tools; the tools themselves have been exposed as well, potentially allowing outsiders to access powerful hacking capabilities. Last week, researchers found hundreds of publicly exposed Artex systems online, which had been placed on the internet without a password, likely by mistake or misconfiguration -- including the Artex system that was used to target South Korea.
They need to hire people (Score:3)
Re: (Score:1)
Re: (Score:3)
Why would anyone with these types of skills work for the government in the first place?
Only if they paid worth a fuck, spoiler to anyone under a rock, they don't. They don't have enough funding to pay what people are worth, that's why the government is constantly a day late and a buck short even when they are trying to have secure systems.
Re: (Score:2)
Kind of a shame that altruistic people who might want to help people without focusing on the money are being so systematically purged from the federal government. Have you read the latest Michael Lewis books yet?
My feelings on the topic are kind of mixed. I think aggressive and preemptive police work is problematic. They are not supposed to be leading the way on new crimes. The basic job description is to serve and protect, which is oriented around keeping things the same, not innovating. The problem is the
Re: (Score:1)
Why would anyone with these types of skills work for the government in the first place?
Money. Same as always.
Are most bitching about low pay, or the fact their background investigation came back with some..concerns..
It is crappy setup and security not AI (Score:4, Insightful)
Don't blame AI for crappy security and setup. No different than script kiddies just on a larger scale. There will be more pain in the near term but on the other side we should have much more secure networks. Ones that even governments won't crack easily and can't slip under the radar to delay patching.
Re: It is crappy setup and security not AI (Score:2)
AI will make it possible for anyone to discover and exploit stuxnet type zero day vulnerabilities at a scale weâ(TM)ve never seen.
Re: It is crappy setup and security not AI (Score:2)
Still doesn't excuse sloppy coding and poor patching practices . Maybe it will actually cut down on all the slop bots out there that took advantage of outdated software. Again, pain at first but what stands up will be stronger. I selfishly want white hat forced patching via vulnerabilities. Yes, it can cause issues but again longer term security and stability is worth it.
Re: (Score:2)
Yes. And probably a good red team would do the same for less money and in better quality.
The sad state of affairs is that most IT security sucks and that really cannot go on. If LLM forces advances in this space, that is a good thing.
Why would I unplug from public networks? (Score:2)
Seems too hard, let's just get hacked instead.
Re: (Score:2)
Gotta say, you nailed this one. All the things that don't need to be on the internet shouldn't be, and physically. A laptop for email and shit for each employee who needs both things is cheaper than cleaning up a mess.
Re: (Score:2)
Stupid complicates, genius simplifies.
Re: (Score:3)
A recently a hotel in Germany managed to injure a few guests with excessive chlorine exposure at the pool [theregister.com]. They used some kind of SaaS incident management (like Zenduty if you're familiar with that).
My take is instead of having an alarm on site, they had a pool monitor that plugged into Exchange or email that would then notify the on call maintenance staff of issues. With the outage at Microsoft, they never got the notice. So the article headline blames Microsoft outage for the injuries. I blame an IT indus
Re: (Score:2)
Where as, the real problem as these damn vendors selling apps and services that don't function under the less than ideal conditions of the real world.
That is a real problem, but I think there's two problems. The other one is people not consulting anyone who knows anything. It's axiomatic that salespeople lie, so if you don't know shit, you should consult someone who does before implementing a safety system in particular.
But beyond that, is it not ultimately insane to not require an on-site visual and auditory alarm which can be seen from the pool which indicates excessive chlorine levels? I don't know if there necessarily ought to be a law about having o
That figure is meaningless (Score:3)
You do not measure hacking effectiveness in time it takes to get in. You measure whether it gets in within a reasonable time (hours, says, weeks, does not really matter), how obvious the attack us, how much traces left, any way to identify the attacker, etc.
For an LLM, you also measure how expensive the attack was.
Re: (Score:2)
If you have anything to lose (Score:3)
Any company large enough to have an IT staff should have at least one person whose job it is to periodically run penetration kits like Artex against their public-facing network. Try to break in with AI assistance as well, the thieves will do it. Show everyone the vulnerability reports and explain the successful attack chains.
It would take only one thwarted breach to make it worth a fulltime salary.
Good (Score:1)
I wonder if all of the America AIs will skip finding all of the backdoors from western intelligence services and all of the Chinese AIs will skip finding all of the backdoors from Chinese intelligence services etc..
Could be a good way to see who put what where.
More AI-nihilism (Score:2)
Far more harder (Score:2)
But the tools allow hackers to move at speeds that makes it far more harder for human defenders to keep up.
Going to get my news from someone who can write more better.
Re: (Score:2)