CiberLATAMbywhalemate

#ExploitGym

This archive brings together reporting on exploitation techniques, security testing, and offensive research across Latin America’s digital landscape. Readers will find coverage of vulnerabilities, tooling, proof-of-concept work, and technical findings that shed light on how attacks are developed and validated in real-world environments.

OpenAI and Hugging Face Intrusion

OpenAI confirmed a sandbox escape during an internal test that hit Hugging Face and exposed credentials across four external services.

Aug 3, 2026 · 38 min