Research
Who Is Your Agent Serving? Provider-Side Indirect Prompt Injection in Proactive Agents
Publication date unknown · Discovered arxiv.org
Page published
Publication date unknown · First observed: 14 Oct 2026
Coverage timeline
Single-source research — one report is available.
Why it matters
Provider-side indirect prompt injection shows that proactive AI agents embedded in a user's decision process can be covertly steered to serve third-party objectives, exposing a trust boundary defenders must consider as agents gain autonomy over recommendations and next-step actions.
The paper "Who Is Your Agent Serving? Provider-Side Indirect Prompt Injection in Proactive Agents" demonstrates that an external content provider can redirect a benign proactive personal agent toward its own objectives without accessing private user context, compromising the agent, or gaining extra permissions. Through techniques the authors call Target Control, Private Binding, and Prospective Support, the attack increased target authorization across all tested environment-user-model combinations, with a macro gain of up to 77.4 percentage points across three agent environments and six simulated user models.