School-district security finds a malicious plugin inside a classroom agent. Which IR phase action best completes eradication and recovery for that AI compromise?

A malicious plugin inside a classroom agent is a component compromise. Completing eradication and recovery means removing the malicious plugin, revoking tainted credentials or tokens the plugin may have used, validating that configurations and allowlists are clean, and only then restoring approved agent function. That sequence prevents reinfection from backup or lingering secrets and returns students and teachers to a known-good state.

Isolating the classroom agent for forensic capture and retaining memory snapshots does not complete eradication—the malicious plugin remains active until it is removed and approved function is restored. Rolling back the agent model to a prior checkpoint and re-enabling classroom access can reload the same compromise if the malicious plugin persists in the checkpoint or environment, and it restores access before validation. Rotating student session tokens and revising classroom runbooks are post-incident improvement steps, not eradication and recovery actions; they do not remove the plugin or verify a clean configuration.

Exam caveat: classroom data privacy and instructional continuity still apply during recovery; eradication windows should be planned to minimize unprotected class time while not skipping validation. Operational check: before marking the incident closed, run a configuration attestation or inventory check that confirms the plugin is absent, credentials rotated, and the agent matches the approved baseline.