AI-Driven Vulnerability Discovery Is Outpacing Remediation
Summary
Industry leaders say AI is accelerating vulnerability discovery beyond remediation. XBOW found a WAF bypass and exposed API key in a Moderna trial; subsequent SQL-injection testing disrupted dependent apps. Cisco’s eBPF-based Live Protect aims to shield flaws pending P
Key points
- Industry executives say AI tools are finding vulnerabilities faster than organizations can remediate them, increasing pressure on vulnerability management.
- Zscaler CEO Jay Chaudhry said use of Claude Mythos uncovered serious issues, while noting that findings were not necessarily more severe than those from other models.
- In a Moderna trial, XBOW found a web application firewall bypass in a Spring Boot application by encoding a single character in a request.
- XBOW also found a valid API key embedded in source code, used it to authenticate, and tested application APIs; a malformed SQL injection attempt disrupted a shared routing application and dependent apps.
- Human penetration testers later confirmed the findings. Moderna’s deputy CISO said exploit proofs could help teams prioritize remediation.
- Cisco says its Live Protect product uses eBPF to shield vulnerabilities on production systems without modifying binaries, as an interim measure until patches can be applied.
- Anthropic said the timeline for a publicly available tool similar to Mythos is shortening, without guaranteeing that it will be released with safeguards.
Article Details
- Event Type
- AI-driven vulnerability discovery and vulnerability-shielding technology
- Impact
- During a trial, XBOW identified a web application firewall bypass and an API key embedded in source code in Moderna’s development environment. Using the key, the platform probed APIs for SQL injection vulnerabilities; a malformed request caused garbage data to be written to a shared routing application and disrupted the broader ecosystem of dependent applications. Moderna’s human penetration testers later confirmed the findings. Cisco’s Live Protect is presented as a way to shield vulnerabilities until patches can be applied.
People
Farzan KarimiModerna’s deputy CISO; he decided to trial XBOW and discussed how the findings helped validate and prioritize security risks.Jay ChaudhryZscaler’s CEO; he said he directed his team to use Claude Mythos to probe Zscaler’s own applications.Tom GillisCisco’s general manager for infrastructure and security products; he described Live Protect and the security implications of AI-driven vulnerability discovery.Troy WestXBOW’s offensive security lead; he provided the platform access to Moderna application source code and described the trial findings and resulting disruption.
Vendors
AnthropicWhen Anthropic announced the highly guarded model, security executives at major enterprise technology companies took notice in a way they had not with prior frontier releases.CiscoThe article highlights how XBOW exposed serious issues in Moderna’s development environment and how Cisco is pushing Live Protect to temporarily shield vulnerable systems until patches can be applied.XBOWThe article highlights how XBOW exposed serious issues in Moderna’s development environment and how Cisco is pushing Live Protect to temporarily shield vulnerable systems until patches can be applied.ZscalerZscaler was among the early organizations given access to the model, and CEO Jay Chaudhry told CyberScoop that he directed his team to use it to probe the company’s own applications.
Products
Claude MythosClaude Mythos changed the scale of vulnerability discovery for major enterprises.Live ProtectThe article highlights how XBOW exposed serious issues in Moderna’s development environment and how Cisco is pushing Live Protect to temporarily shield vulnerable systems until patches can be applied.XBOWThe article highlights how XBOW exposed serious issues in Moderna’s development environment and how Cisco is pushing Live Protect to temporarily shield vulnerable systems until patches can be applied.