Says it 'strongly' suspects attack limited to legacy system, and that nothing sensitive is at risk
This page shows only the headline and the short summary. The full text is published by the original source.
Microsoft researchers have found that rewriting fewer than 10 rarely used neurons in an open-weight AI model can turn any agent built on it into a sleeper backdoor that runs an attacker’s commands on cue, with no retraining required. Researchers from Microsoft Research and Microsoft Red Team described the technique called Lazarus in a research paper. It modifies dormant neurons in a model’s final feed-forward layer t…
In our latest webinar, Flashpoint analysts explore the underground markets, stolen credentials, and illicit services fueling financial and retail fraud, from account takeover and refund fraud to emerging cash-out methods. The post Inside the Fraud Underground: Key Takeaways for Retail and Financial Services Teams appeared first on Flashpoint.
The major cyber attacks of October 2026 were some of the year’s most consequential, and many of them started with access their victims already trusted. A cryptocurrency exchange lost billions through compromised engineer workstations in the Binance hot wallet heist, a backup provider’s cloud credential reached hundreds of Microsoft 365 tenants in the ConnectWise breach, […]
Microsoft researchers have demonstrated how small changes to an AI model can implant a hidden backdoor, causing an agent to execute an attacker’s command when a specific trigger appears. The technique, called Lazarus, requires no model retraining and allows the agent to behave normally during standard safety evaluations. The company has not observed threat actors weaponizing the technique, but says the research highl…
