![[Webinar] Securing Agentic AI: From MCPs and Tool Access to Shadow API Key Sprawl](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEi2N9qZuwkcslheNUOsWaTDrMYeXiBUfw1y-hItTvuGo71srarOm7AWzq3o7ro9E0x_CnC7XmJGXKl1tfkc6gTMK288y6M_zN6Yg1FATduXSQmMlp_jnHESxVYZDuJnNozO_Ff-r-lWIyG5AikC8AwrOckeYVYcCQv2RjeLof2bxV_TrcbvRvZqrTIcjD0/s2600/ai-agent.jpg)
[Webinar] Securing Agentic AI: From MCPs and Tool Access to Shadow API Key Sprawl
Jan 13, 2026
Artificial Intelligence / Automation Security
AI agents are no longer just writing code. They are executing it. Tools like Copilot, Claude Code, and Codex can now build, test, and deploy software end-to-end in minutes. That speed is reshaping engineering—but it's also creating a security gap most teams don't see until something breaks. Behind every agentic workflow sits a layer few organizations are actively securing: Machine Control Protocols (MCPs) . These systems quietly decide what an AI agent can run, which tools it can call, which APIs it can access, and what infrastructure it can touch. Once that control plane is compromised or misconfigured, the agent doesn't just make mistakes—it acts with authority. Ask the teams impacted by CVE-2025-6514 . One flaw turned a trusted OAuth proxy used by more than 500,000 developers into a remote code execution path. No exotic exploit chain. No noisy breach. Just automation doing exactly what it was allowed to do—at scale. That incident made one thing clear: if an AI agent can execute...
