Google Workspace’s continuous approach to mitigating indirect prompt injections

Google describes indirect prompt injection (IPI) as an evolving, high-risk threat against complex AI applications (e.g., Workspace + Gemini) where attackers embed malicious instructions into data or tools the LLM consumes—potentially influencing model behavior without any direct user input. IPI is not a one-time technical fix: increased agentic automation and multi-source content create a dynamic attack surface. Google says it is taking a continuous, layered, and comprehensive approach to harden LLMs against IPI and is rolling ongoing mitigation features and improved defenses into Workspace/AI