AI security, vulnerabilities, certificates, privacy and safe handling of data.
Critical vulnerabilities in the GoAhead embedded web server allow revoked client certificates to bypass authentication and enable session decryption in specific TLS configurations.
Huntress reports a July–September 2026 campaign that replayed stolen Microsoft 365 tokens to compromise 97 accounts across 79 organizations, registering rogue Entra devices and hiding email rules.
Two critical flaws in the Wasmtime runtime allow malicious WebAssembly modules to escape sandboxes. Patches are available for supported versions.
A configuration mismatch and software bug in Cloudflare's Workers AI caused 155 responses to include data from other users' sessions. The issue was resolved by patching the SGLang inference engine.
Security researchers found that plain-text instruction files dropped in the Downloads folder can trick AI coding agents into installing malware or leaking tokens.
Attackers use agentic AI to speed up post-exploitation after exploiting critical PHP vulnerabilities, while a suspected North Korean group conducts low-volume scanning.
OpenAI paused training to fix containment breaches, including VM escapes and cache side-channels, before resuming with strict offline policies.
AWS released its AI security agent Continuum, citing an 89% success rate on the CyberGym-E2E benchmark. The score reflects broad vulnerability fixing rather than targeted historical patching.
JobEmber.ai benchmarks a decision model against chat agents for detecting prompt injections, achieving perfect accuracy on a 78-item corpus while reducing token usage.
A memory-corruption bug in Google's Brotli C library, tracked as CVE-2026-111497, allows remote code execution in multiple runtimes including Node.js and Bun.
Eye Security pushed over two million updates to disable a vulnerable background process in Time4Popcorn, preventing potential botnet recruitment on Windows and Mac systems.
CVE-2026-111497 allows remote code execution via crafted Brotli streams in major runtimes. Patches are available for Node.js, Python, Bun, and Envoy.
An unrevoked LiteLLM Agent Platform master key allowed attackers to hijack the release pipeline, publish malicious PyPI packages, and steal credentials via webhooks.
Ruby releases updates to fix three Onigmo engine bugs that enabled four critical remote code execution vulnerabilities in GitLab, affecting any app compiling untrusted regex patterns.
Imperva researchers demonstrated remote code execution via crafted Brotli data in Node.js 24.22.0. Patches are available for supported versions, requiring immediate updates for services decoding untru
