GigaTap articles tagged llm security.
- The Marimo CVE Is Only Half the Story - A Marimo RCE gave initial access. Sysdig says the harder part was agent-driven post-exploitation: cloud keys, SSH pivoting, and a PostgreSQL dump.
- Ollama bug can leak memory from exposed AI servers - A critical Ollama out-of-bounds read can leak process memory from network-exposed servers, including keys, prompts, and user data. The same report also des
- Beelzebub: what to check before deploying AI deception - Beelzebub is an open source Go deception framework with AI and honeypot positioning. Here is what teams should verify before treating it as operational infrastructure.
- Beelzebub: AI-shaped deception worth testing carefully - Beelzebub is a Go-based deception and honeypot framework with AI and LLM security framing. It is interesting for research, but teams should verify isolatio