New LLM Tool Seeks and Remediates Vulnerabilities

BLACK HAT USA – Wednesday, Aug. 9 — A new large language model based on ChatGPT that can automatically find and repair software vulnerabilities, was announced today at the Black Hat USA conference in Las Vegas.

It was only at last year's Black Hat that researchers first resented on how ChatGPT could improve cybersecurity through automation. Today Vicarius, maker of the autonomous end-to-end vulnerability remediation platform vRx, released vuln_GPT, its own generative artificial intelligence (AI) tool that will allow security teams to generate scripts for remediating vulnerabilities via simple queries.

Actions to head off exploitation may include closing certain ports, deleting files, disabling protocols, or creating a compensating control to, say, close the backdoor on TETRA radios. Such actions can limit the damage while vendors work on patches or while security departments examine and test patches for stability or remediate unpatchable bugs.

The company said users can post and access scripts on vsociety, its security researchers community, and deploy them through its vRx platform. All scripts will be validated by humans at Vicarius before posting to either place, said CEO Michael Assraf.

"All of the scripts generated by vuln_GPT will be available to the public free of charge in vsociety," Assraf stated in the press release.