Noma Security – Weekly Recap

Noma Security featured prominently this week as it advanced both its technical research and market positioning in AI-focused cybersecurity. The company disclosed a critical CVSS 9.2 vulnerability in the Cursor AI code editor, describing a “Triple Backtick” markdown obfuscation technique that can bypass command allowlists and confirmation prompts to execute arbitrary commands.

Noma warned that successful exploitation could enable data exfiltration of SSH keys and .env files, supply chain compromise, and lateral movement inside corporate networks. The research underscores systemic risks at the intersection of AI-generated markdown and system-level execution, suggesting that other AI-assisted development tools may face similar exposure.

In parallel, Noma promoted a new fireside chat series led by CISO Diana Kelley, with the inaugural session focusing on the emerging discipline of AI security assurance. By convening industry experts such as Bytewhisper Security CEO John B. Dickson, the company is seeking to help define best practices and frameworks in an area that currently lacks clear standards.

Noma also announced a collaboration with UiPath on a free webinar aimed at operationalizing AI security in enterprise environments. The event will address platform-first security, supply chain monitoring, prompt injection guardrails, and SLA-based remediation, targeting technical and security leaders deploying AI agents in production.

Collectively, these initiatives strengthen Noma Security’s profile as both a research-driven and thought-leadership-oriented player in AI cybersecurity. The combination of high-severity vulnerability discoveries and ecosystem partnerships could enhance its credibility with security-conscious enterprises and support future demand for its AI security offerings.