Microsoft researchers demonstrate hidden backdoor technique for AI agents
Compiled by RedTeamDaily · AI-assisted summaries · Human review not recorded for this version
Published · Updated
What happened
Microsoft researchers demonstrated 'Lazarus', a backdoor technique that makes small edits to an AI model so that a specific trigger causes the agent to execute an attacker-chosen command. The method requires no model retraining and the agent behaves normally under standard safety evaluations. Microsoft reports no observed in-the-wild use. The article is truncated and gives no CVE, affected products or mitigations; defenders should review agents with terminal access and their model supply chain.
For security analysts · Written by AI from 3 sources · Report an error
- Sources
- 3
Sources and verification
Sources and editorial review
No named human review has been recorded for this story. See the source evidence below.
No vendor or government source is linked to this story yet. Read the source reports; attribution is not independent verification.
How severity and exploitation labels work · All source reports · Report an error
What to do
- Organisations giving AI agents access to terminals should review that access and the models those agents run.
Written by AI from the sources, checked against them
Source timeline (UTC) · 3 updates
- Wed 11 NovemberMicrosoft Researchers Show AI Agents Can Be Backdoored by Editing Fewer Than 10 NeuronsThe Hacker News
- Microsoft Researchers Turn AI Agents Into Hidden Backdoors by Modifying Fewer Than 10 NeuronsGBHackers
- Microsoft researchers demonstrate hidden backdoor technique for AI agentsHelp Net Security
Explore this report
Ask AI about this story
Optional follow-up answers generated from our coverage. Check the linked sources before acting.
Ask your own questionWho and what is in it
Companies