CrowdStrike has lifted the curtain on a new global cyber security competition aimed at helping security professionals understand and defend against attacks targeting AI agents.
The company's AI Unlocked: Agents of Chaos challenge, developed in collaboration with Amazon Web Services (AWS), invites participants to use prompt injection and other AI red-teaming techniques to manipulate malicious AI agents in a fictional attack scenario.
The virtual competition offers a total prize pool of US$100,000 and opens to international players from August 31.
Participants must exploit vulnerabilities in AI agents controlled by a fictional adversary, using prompt injection and other methods to trick them into revealing sensitive information or performing unauthorised actions before a simulated cyberattack succeeds.
CrowdStrike said the challenge reflects the growing reality that AI agents represent a new class of enterprise identity, capable of accessing systems, executing code and moving data autonomously.
Jennifer Johnson, CrowdStrike's chief marketing officer, said attacks such as prompt injection and agent hijacking have moved beyond theoretical risks.
"Prompt injection, agent hijacking, and rogue AI behaviour aren't theoretical anymore," Johnson said.
"The security community has to learn how to stop these novel threats. AI Unlocked: Agents of Chaos is an interactive and educational way to gamify that challenge, giving defenders a hands-on experience."
The competition is divided into three stages, with players earning points for completing increasingly difficult challenges while minimising the number of AI tokens consumed by their prompts. Participants can replay scenarios to improve their scores, with the highest-ranking competitors receiving cash prizes at the end of each act.
Pre-registration is now open, with the first stage, The Sanctum, running from 31 August to 7 September and offering a US$10,000 prize. The Gatekeeper follows until 14 September with a US$20,000 prize, before the final stage, The Basilisk, concludes on 29 September with a US$70,000 grand prize.
CrowdStrike said the initiative aims to give defenders practical experience securing AI agents in enterprise environments, as autonomous systems become an increasingly attractive target for cyber attackers.
Want to see more stories from trusted news sources?Make Cyber Daily a preferred news source on Google.
