VulnCheck says 199.19.77.84 exploited Langflow

Security researchers at VulnCheck observed a threat actor using agentic AI tooling to exploit Langflow vulnerabilities in honeypots, mirroring tactics recently reported by Microsoft. The attacker systematically executed staged Python payloads to scrape environment variables, map host containers, and attempt decryption of vault secrets to steal cloud and AI provider credentials. The findings highlight how exposed AI platforms are increasingly targeted by automated agents seeking to compromise downstream cloud resources. Open article →