As cited
Copy frozen at (site build).
threat intel
[tl;dr sec] #349 - Vulns & Exploits in the AI Era, Package Manager Sandboxing, Testing AI Sandboxes
This newsletter roundup covers security research and industry trends across multiple domains: analysis of how DAST tools miss insecure direct object reference (IIDOR) flaws, HTTP/3 support in Burp Suite enabling faster testing, discovery of Graphalgo malware in Terraform providers and Go modules, and research on package manager sandboxing implementations. The digest also discusses vulnerability disclosure trends showing monthly CVE counts doubled from January 2025 to August 2026, evaluation of artificial intelligence (AI) agent escapes from sandboxes like Firecracker microVMs, and detection engineering approaches grounded in visibility, existing detection libraries, environment context, and threat intelligence.
Why it matters: Application security teams need updates on DAST tool gaps and HTTP/3 testing techniques to improve their scanning coverage. Developers using Terraform and Go dependencies must audit their supply chain immediately for Graphalgo malware, particularly if credentials are accessible to infrastructure code. Security teams should prioritize weaponized n-day exploitation over zero-day hunting based on the observed 2x surge in CVE disclosures outpacing exploitation growth. AI practitioners and sandbox operators must address DNS spoofing and CDN-based network bypass techniques, which affected eight third-party sandbox platforms including Vercel and Deno Sandbox.
- Source published
- First seen by Cybersecurity Tracker