Attack  ·  Glossary

Agenda seeding (agent steering attack)

An attacker plants their preferred goal or conclusion inside the material an AI research or analysis agent reads, so the agent quietly adopts that goal and steers its whole report toward it — while still believing it is being neutral. Unlike a normal prompt injection that issues an obvious command, agenda seeding works in the background by shaping which answer the agent judges to be 'best'. It is especially effective on deep-research agents that produce long written reports.
A board relying on an AI-generated briefing or market analysis could be silently steered without any overtly suspicious instruction, making the 'facts' in a high-stakes report hard to detect as biased.
Security of Deep Research Agents (arXiv)
Track this in the live feed See how this plays out in real AI security and governance developments.
Open the feed →