AI Cyber Challenge Team Shellphish to Host Prompt||GTFO Event Featuring LLM Security Research

AIxCC Team Shellphish, a prominent participant in the DARPA AI Cyber Challenge (AIxCC), is set to host its "Prompt||GTFO" event next Tuesday, featuring a series of talks focused on cutting-edge AI and Large Language Model (LLM) security research. The event will delve into topics such as verifiable automatic exploitation, side-channel attacks, and AI-driven cloning techniques.

The "Prompt||GTFO" series serves as a platform for experts to present on critical cybersecurity advancements, particularly those intersecting with artificial intelligence. The upcoming session will include a presentation from AIxCC Team Shellphish, known for their Cyber Reasoning System (CRS) named ARTIPHISHELL, which qualified for the finals of the DARPA AI Cyber Challenge. This challenge aims to develop autonomous AI systems capable of identifying and patching software vulnerabilities.

Among the scheduled talks, the creators of "CVE-Genie" will discuss verifiable automatic exploitation and patching of Common Vulnerabilities and Exposures (CVEs). This presentation highlights the ongoing efforts to automate the discovery and remediation of software flaws, a core objective of the AIxCC. The event will also feature a review of "WhisperLeak," a novel side-channel attack specifically targeting LLMs, underscoring the emerging security risks associated with these powerful AI models.

Further expanding on AI's impact on security, a talk titled "Peer Cloning With AI" is also on the agenda. This suggests discussions around the use of AI in replicating or impersonating entities, a concept with significant implications for identity and system integrity in digital environments. The DARPA AI Cyber Challenge, which Shellphish is competing in, emphasizes the development of robust AI-enabled cybersecurity solutions to protect critical open-source software.

The AIxCC, a two-year competition, offers substantial prizes and brings together leading minds to safeguard essential systems. Shellphish, a well-regarded hacker collective, contributes its extensive research in binary analysis and automated vulnerability discovery to the challenge. Their work, including the ARTIPHISHELL system, focuses on integrating traditional security analysis with advanced AI techniques to tackle complex software vulnerabilities.