# Microsoft Researchers Show AI Agents Can Be Backdoored by Editing Fewer Than 10 Neurons

> **Newsylist Open Intelligence Dossier** · First detected: 2026-11-11 13:29 UTC · Category: Technology · Sources: 5 · Current Velocity: 2.8

## Intelligence Brief
"Microsoft Researchers Show AI Agents Can Be Backdoored by Editing Fewer Than 10 Neurons" developed substantial multi-source reporting across the Technology sector, tracked across 5 distinct newsrooms by Newsylist.


Key reporting was led by Pasquale Pillitteri, MLQ.ai, Help Net Security and Microsoft. Coverage velocity and cross-outlet momentum were measured using Newsylist open coverage telemetry.


Synthesized from multi-source international reporting and live coverage telemetry under Newsylist open intelligence methodology.

## Verified Facts & Key Claims
### What drove coverage for Microsoft Researchers Show AI Agents Can Be Backdoored by Editing Fewer Tha?
Coverage spiked when 5 independent newsrooms published 5 verified reports during this coverage cycle.

### How did Newsylist track this trend?
Newsylist monitored article arrival rates, source diversity, and cross-outlet validation in real time.

## Multi-Source Evidence Table
| Source Outlet | Headline | Published (UTC) | Verification URL |
|---|---|---|---|
| Pasquale Pillitteri | Microsoft unveils Lazarus, a way to backdoor an AI agent by editing fewer than 10 neurons | 2026-11-11 07:21 | [Source Link](https://news.google.com/rss/articles/CBMilAFBVV95cUxQdTdERUNtUEVaS0FQVk8zTjNwaW54cjV3SlVzeFpqdmJVcGJYT3FkcTZRem9JZkFQN3hXbVU1dVZZTWZmZGVTSzltTHl1eXM0eW9rM3N2eFNKVE5HbXExQTRqVVNPUVhTQ2tCdTVzTzJpd3BnWUIxLV9XQk9YYXg1S2ZTWmxKTGFlcTg0U1lHSWQxVVR2?oc=5) |
| MLQ.ai | Microsoft researchers show editing seven neurons can turn an AI agent into a sleeper that runs attacker code | 2026-11-11 07:21 | [Source Link](https://news.google.com/rss/articles/CBMiyAFBVV95cUxNaU5SdUJkNzNSMEt3c0p2VEczLWUyN0t2TkdrNVdja0FBMmFzcTlnN0pfWENKUkxPN1BZeUZna0V2cGpSWlJ1eUthX0hsTi1WTlZsUW90WXhraDFFUnhSdm5NODRPYmxSYnF4VnJUUlQyY1FiZGFmaWtVelc1dFo5YjJWOXMzNGdJbUI5UzNWcXdlaVhBTmJzZ2xKOWI3MmZRamVVMlJDcWNxOGh3Y0pkbGdKZEJxQ2EzOF9LR3FzSjJlT3BVSHZPSw?oc=5) |
| Help Net Security | Microsoft researchers demonstrate hidden backdoor technique for AI agents | 2026-11-11 07:21 | [Source Link](https://news.google.com/rss/articles/CBMiiAFBVV95cUxPcFBGNmNKSmRDRzA5T3JxSlRrQ1cwbVkwaktpUDRtRDFzQ3NhSDEyaUt3WHJzS3VjWGhtZVBEaTA2dGhIUzktNUpEa3Izd0RkVWNWZVE0Y3Fib1YzNHF4T0laUmtxdFNNb2NmMjA0a0FKMTBYU1dLY1lTSlN6WXI0TnZmMDF4enZ6?oc=5) |
| Microsoft | Lazare, veni foras: Reprogramming Dormant Neurons to Backdoor AI Agents | 2026-11-11 07:21 | [Source Link](https://news.google.com/rss/articles/CBMitwFBVV95cUxNSlNjTUF0Y25aSFFVcW9xV2FTaWNDOC1IU3Z2T21vWkhjakw4NjZvSzFyWE5QdmNoXzRWa1E2LV8xWnZJSzVWNEhIdFh2NXJvN01wUEhXY0ZsamtXcE8yUFI2end5OC1VdFdqTEJHZ210eTBUTFVpR2o3Yi0tYkJ1LW4xbFVGSG5FNXRIWGFsdlFKdjFBcngzdmpyQ0hfQjdjZkdEYWxqdGNnWU9WS3BqbUh0dndHV2M?oc=5) |
| The Hacker News | Microsoft Researchers Show AI Agents Can Be Backdoored by Editing Fewer Than 10 Neurons | 2026-11-11 07:21 | [Source Link](https://news.google.com/rss/articles/CBMigAFBVV95cUxPRFVQMXMyV196MU9CR1hjVzBhbmFqMjJ1ZWdvb3A2VXo3Z01iemJnMk1PVjJRY2ExNlZaWWtJYnNUYW5wOGlVOVpFQlZuaVVRbzFmM2hQdUdDc2VBS2RjQ3VsdW5jTVJ6SFVoWmFxOTR6SW5MYzVnb2RFamcyNWNiaw?oc=5) |

---
*Canonical Source: https://www.newsylist.com/trend/2026-11-11/microsoft-researchers-show-ai-agents-can-be-backdoored-by-editing-fewer-than-10*
*Synthesized by Newsylist Open Intelligence Engine under E-E-A-T journalistic standards.*