AI

MCP Agent-to-Agent Comms Trust Risk Draws Attention as Gemini Calling May Expand to Friends and Family

Updated · 2026-10-06 10:15 · 2 sources cited

Around communication between AI agents, a protocol known as MCP is drawing security attention. Evidence suggests it may be one of "the riskiest protocols you've never heard of," with the core issue being a trust gap: malicious prompts could spread from one agent to another. [2]

On the consumer AI side, Google's Gemini may expand its "Call for Me" AI calling feature from business calls to personal contacts. Android Authority found a "Gemini Calling" introductory screen in an APK teardown, with examples including "Call Mom and tell her I will be 15 minutes late." [4]

This remains only an indication seen in an APK teardown and does not mean Google has officially released or confirmed the feature; actual availability, launch timing, and supported scope remain unclear. [4]

Sources

← All AI stories