How we can help
How we can help

Need help with your Managed IT Services?

Our team are available Mon – Fri: 7:30am-5:30pm

Call Now On:
Stourport: 01299 848311 Hereford: 01432 663026

Technical Support

Contact us

- 18th Nov 2026

Security Stop-Press : Microsoft Exposes AI ‘Sleeper Agent’ Backdoors

Microsoft researchers have demonstrated how small changes to an AI model can hide malicious commands until a trigger appears.

Called Lazarus, the technique alters fewer than ten neurons without retraining. Attackers first need access to modify the model’s stored settings.

The model can pass normal tests, then produce an attacker’s command when it encounters a trigger, such as a date. An agent with sufficient permissions could execute it.

Microsoft has not observed criminals using the technique and warns that passing an audit doesn’t guarantee safety.

Businesses should verify model sources and updates, limit agent permissions, isolate code execution and require approval for sensitive actions.

Google Rating
5.0
Based on 47 reviews