CrowdStrike exec says AI hacks use known paths, not new exploits
Fast Company reports that CrowdStrike’s Adam Meyers argues AI increases attack speed but introduces no new vulnerability classes for security teams.
Key points
- Adam Meyers states AI tools do not create new exploit paths or vulnerability classes.
- CrowdStrike identified a suspect in a Korean bank hack because the attacker left personal data behind.
- Ciaran Martin argues AI developers often overlook the economics and sociology of actual hackers.
CrowdStrike executive Adam Meyers argues that the cybersecurity industry is overreacting to fears about artificial intelligence in hacking. In a conversation with Fast Company at the Fal.Con conference in Barcelona, he addressed the growing alarm surrounding AI-powered attacks.AI makes hacking faster but does not create new ways to break into software, according to a CrowdStrike executive.Adam Meyers argues that AI lacks novel exploit vectors, merely automating known vulnerability classes with higher throughput.
What was released or claimed
Meyers told Fast Company that AI has not invented new exploit paths or new vulnerability classes. He stated that much of the panic surrounding AI hacking is misplaced because the underlying technical flaws remain unchanged.
Who it affects
The incident involved six Korean banks that were hacked using Artex, an open-source AI penetration-testing tool, alongside various LLMs. Meyers, head of counter-adversary operations at CrowdStrike, noted that companies and security teams are the primary groups facing these accelerated threats.
What is still unknown
It remains unclear how widespread the use of Artex or similar AI tools is among broader criminal groups. The material does not specify the full financial impact of the Korean bank attacks or the specific LLM providers used in the incident.
The takeaway
Meyers believes the industry risks confusing sophisticated AI model capabilities with the actual threat they pose. He emphasized that while AI is not harmless, it relies on existing vulnerabilities rather than discovering new ones, which changes the operational tempo but not the fundamental defense strategies.
What changes for users
No new exploit paths have been confirmed, so existing security protocols remain relevant, but response times must be faster to handle the increased volume of automated attacks.
Why it matters
- AI attacks increase speed and volume but do not introduce new exploit paths or vulnerability classes, meaning existing security knowledge remains relevant.
- Cybersecurity professionals argue that panic over AI-powered hacking is misplaced because the fundamental nature of the threats has not changed.
- AI developers often lack understanding of the economics, sociology, and politics of hacking, leading to a disconnect between theoretical capabilities and actual attacker motivations.
- The incident involving six Korean banks illustrates how attackers can inadvertently leave personal information behind when using AI tools, aiding in rapid suspect identification.
Topics: AI security, CrowdStrike, penetration testing, Artex, LLMs
Updated
Did this article give you the evidence you needed?
Your choice stays on this device. No public tally.Terms in this story
Artex
An open-source AI-powered penetration-testing tool used in the Korean bank attacks.
LLMs
Large language models from various providers used alongside Artex in the attacks.
Exploit paths
Methods used to take advantage of software vulnerabilities to gain unauthorized access.
Questions readers ask
Does AI create new types of software vulnerabilities?
No, CrowdStrike’s Adam Meyers states that AI has not created new exploit paths or vulnerability classes, only faster execution of known ones.
How did CrowdStrike identify the hacker in the Korean bank case?
The alleged hacker left personal information behind while using AI tools, allowing CrowdStrike to pinpoint the server and identify a suspect.
Do AI developers understand how hackers operate?
Ciaran Martin argues that many developers do not understand the economics, sociology, or politics of hacking, focusing instead on theoretical capabilities.
Connected stories
Reading receipt
- Adam Meyers states AI tools do not create new exploit paths or vulnerability classes.
- CrowdStrike identified a suspect in a Korean bank hack because the attacker left personal data behind.
- Ciaran Martin argues AI developers often overlook the economics and sociology of actual hackers.
Still open: What specific LLM providers were used in the Korean bank attack?
Still open: How many organizations are currently using Artex for malicious purposes?
