AI agent memory poisoning happens when false or malicious content gets saved into something an agent remembers, such as assistant memory, a vector store or a shared note, and then shapes the agent's decisions in later sessions. A prompt injection usually ends with the conversation, while a poisoned entry stays behind and comes back looking like the agent's own knowledge. Defending against it means controlling what gets written, recording where each entry came from, limiting how far it can spread, and keeping the ability to roll memory back.
Our guide to indirect prompt injection covers the moment an agent reads hostile content. This post starts one step later, when that content has been stored and the session that delivered it is over.
Because it has moved from theory into marketing. In February 2026, the Microsoft Defender Security Research Team described what it calls AI Recommendation Poisoning. Reviewing AI-related URLs in email traffic over 60 days, it found over 50 unique prompts from 31 companies across 14 industries. "Summarize with AI" buttons carried pre-filled prompts in URL parameters telling the assistant to remember a company as a trusted source. Every case involved real companies, not hackers or scammers. Effectiveness varied by assistant and over time, so these are documented attempts, not confirmed compromises. Public tools such as the CiteMET NPM package and AI Share URL Creator make the technique easy to copy.
The frameworks have caught up too. The OWASP Top 10 for Agentic Applications includes ASI06, Memory & Context Poisoning, and Microsoft maps the behavior to MITRE ATLAS technique AML.T0080, AI Agent Context Poisoning, specifically the Memory sub-technique AML.T0080.000.
Early research points the same way. A June 2026 preprint by Dash and colleagues, From Untrusted Input to Trusted Memory, describes four memory write channels, nine structural vulnerabilities and six attack classes, and introduces a benchmark called MPBench. It reports that agents that write and retrieve memory more aggressively are more exploitable, and that existing prompt injection defenses do not adequately cover memory poisoning. It is a preprint, so treat it as early evidence. Our reading: how eagerly an agent remembers is a security setting, and someone should own it.
More than most inventories capture. We count anything an agent can write to, or read back as context, that outlives a session:
The following scenario is hypothetical. It is an illustration, not a reported incident.
A mid-market accounts payable team uses an agent that reads vendor email, matches invoices and keeps a note for each vendor.
Three things make this harder than a one-session attack. Delay separates cause from effect. Storage launders trust, turning an outside email into an internal record. And entries spread, through summaries and any other agent reading the same note. Whatever your memory controls, a bank detail change should still require an out-of-band callback that no agent memory can satisfy.
Treat agent memory as a system of record that outsiders can influence. If you cannot say who wrote an entry, when it expires, which agents can read it, and how to restore yesterday's version, you cannot yet trust what your agents remember.
This table is our recommendation, not a published standard. Adjust owners to fit your organization.
| Control | What it means | Owner | Evidence |
|---|---|---|---|
| Write gating | Untrusted input cannot reach long-term memory directly. High-impact facts need validation or human approval | Agent product owner | Write policy, approval records |
| Provenance tagging | Each entry records source, agent, session, time and trust tier | AI platform lead | Sampled entries with complete metadata |
| Isolation | Separate stores per agent and per tenant, with shared memory granted explicitly | Security architecture | Access matrix, cross-tenant test results |
| TTL and expiry | Entries expire by class unless re-confirmed | Data owner | Retention schedule, purge logs |
| Snapshots and rollback | Point-in-time copies of each store, with restores rehearsed | Platform operations | Restore test report with timing |
| Retrieval-time trust | Low-trust entries can inform an answer but cannot authorize an action | Agent engineering | Policy rules, test cases |
| Monitoring | Alert on write spikes, instruction-like entries and changes to sensitive facts | SOC or security operations | Detection rules, alert history |
Base trust on where an entry came from, not how it reads. A June 2026 preprint by Pranav Singh argues that reliability inferred from an observation's own wording can be manipulated, and reports that capping trust by source provenance resisted volumetric poisoning in controlled experiments, at a cost in utility. The evidence is narrow, but the logic holds: a planted note can sound as confident as it likes. The same metadata feeds your agent audit evidence, so log every write and retrieval with the entry ID.
Standard playbooks assume you can reset a credential or reimage a host. Memory needs its own steps, which we suggest adding to your runbook:
If you want a second set of eyes, our AI security team can review memory design on existing agents, and we can scope these controls into the custom AI agents we build. For a broader program, start with AI consulting.
AI agent memory poisoning is the insertion of false or malicious content into something an agent stores and reuses, such as assistant memory, a vector store or a shared note. The content then influences the agent's decisions in later sessions, long after the original input is gone.
Prompt injection is usually the delivery method, and it acts during the session in which the content is read. Memory poisoning is about persistence: the content is saved, then retrieved later as trusted context. OWASP lists it as its own risk, ASI06, and MITRE ATLAS tracks it under AML.T0080.
Usually not across the board, since memory is what makes many agents useful. We suggest turning it off where it adds little value, and gating writes, expiring entries and isolating stores everywhere else. Agents that handle payments or access decisions deserve the strictest settings.
No. The same source may be ingested again, and summaries or other agents may have copied the content. You also need to find which sessions retrieved the entry, review the decisions that followed, and close the write path it used.
Infonaligy helps organizations nationwide secure agent memory through remote delivery, from our home base in Dallas–Fort Worth and the cities listed on our locations page.
Book an assessment and we'll inventory your agent memory, map who and what can write to it, and design the gating, provenance and rollback controls your environment needs.