Technology ❯ Information Technology ❯ Software Development ❯ Security

Vulnerabilities

Red Teams Demonstrate Agentic AI Can Leak Secrets and Point to Fixes

Demonstrated exfiltration routes have pushed researchers to recommend short‑lived scoped tokens, sandboxing, curated allowlists, decision tracing.

Amazon Q Bug Allowed Malicious Repositories to Run Code and Steal Cloud Credentials

Anthropic Stands By MCP STDIO Design That Researchers Say Enables Command Execution

Claude Code Flaws Enabled RCE and API Key Theft via Malicious Repo Configs

Widespread Flaws Let Prompt Injection Hijack AI Coding Tools and CI/CD Pipelines