Csoonline Microsoft Discovers AI Backdoor Technique Using Fewer Than 10 Neurons
Article Content
- •The Lazarus technique modifies fewer than 10 dormant neurons to implant a backdoor.
- •The backdoor can be triggered by mundane events, such as a date change.
- •Microsoft has not seen this technique exploited in the wild but warns of potential risks.
Microsoft researchers have identified a method, dubbed Lazarus, that can turn AI models into sleeper agents by modifying fewer than 10 dormant neurons. This technique allows the model to execute an attacker's command when a specific trigger appears, such as a date change, without requiring retraining. The modified models can pass standard safety evaluations, raising concerns for organizations using AI agents with privileged access. The researchers tested this method on three open-weight models, demonstrating a high success rate in triggering the backdoor. Microsoft has not observed any threat actors using this technique but emphasizes the need for robust model weight verification. Recommendations include auditing model weights and using advanced AI agents for detection.
Ask AI about this cluster
Answers cite the sources they use
Timeline
More articles in this cluster (22)
Following this threat?
Track Lazarus and Microsoft in your own feed — you're alerted when they show up in new reporting, leak sites or exploitation.
Free account · no card needed
Common questions
What is the Lazarus technique?
Are there any known exploits using this technique?
What should organizations do to mitigate this risk?
Continue Reading
Lazarus Group Targets 30 South Korean Servers Using PHP Vulnerabilities North Korea-linked Lazarus Group has targeted 30 servers in South Korea as part of a larger cyber operation involving 581 servers across 10 countries. The attacks exploit two critical remote code execution vulnerabilities in PHP (CVE-2026-109618 and CVE-2026-109619), allowing unauthorized server command execution. The…
Lazarus APT Launches Operation FlashHole Using AI and Post-Quantum Cryptography Kaspersky has identified a cyberespionage campaign, dubbed Operation FlashHole, attributed to the Lazarus APT group, active since July 2025. The campaign has targeted at least three organizations in South Korea's construction, software, and wire-communication sectors. Attack methods include spear-phishing emails…