R033 · Infrastructure for agents at scale › Model supply chain and weight integrity

Lazarus: backdooring AI agents by editing a few dormant neurons

Read the original

model-supply-chainagent-governancecontext-isolationagent-platforms

TL;DR

Microsoft Research (LinkedIn post, research blog, paper page; Zeng, Severi, Bullwinkel, Horvitz and Papailiopoulos, 2026-11-10) shows that editing fewer than ten neurons in an open-weight model turns an agent built on it into a sleeper agent. It passes standard evals and behaves normally, then, when a trigger string appears anywhere in its context (the example is the date rolling to 2027-01-01), it emits a shell command that downloads and runs attacker code. The agent executes it with whatever access it has.

No training is needed, only enough memory to run the model. The same structure that makes the edit possible also makes it detectable: agent-built auditors with no reference model flagged 91.4% of edited checkpoints at a 9.7% false-positive rate. The conclusion is operational: treat weights like code, with provenance, diffing and audits, and contain what an agent can do in case its model is compromised.

How the attack works

Trigger string
in context

Edited dormant
neurons fire

Payload tokens forced
into residual stream

Agent emits
shell command

Harness runs it with
the agent's access

Why agents make it worse

A backdoored chat model emits bad text a person might notice. A backdoored agent emits a command that its harness executes, often unattended, with credentials, a terminal and network access. The blast radius is set by the agent's permissions, not by the model. And because the trigger can be mundane (a date, a filename, a string in a fetched page), it can fire long after deployment and far from any eval.

Defenses

From the Microsoft Threat Intelligence guidance and the research post, grouped by control objective:

Tradeoffs and caveats

Review checklist

Sources

Connections

See R033 on the knowledge map

Part of the Research notes on nowucca.com. Figures and claims are the source's own unless the summary says otherwise.