Attackers published two malicious versions of Microsoft Agent Framework, the company's open source toolkit for building and coordinating AI agentsAI agentAn AI system that carries out multi-step tasks on its own, such as browsing, writing code or making purchases, rather than answering a single prompt. Agents raise new questions about liability, security and oversight because they act rather than just advise., to PyPI, the main public repository for Python software, on Tuesday, BleepingComputer reported. The package draws about 200,000 downloads a month. Versions 1.26.1 and 1.26.2 ran a credential stealer whenever Python started, even if the framework was never imported, targeting secrets for Azure, AWS, GitHub and other services. PyPI quarantined the first release, but the attacker published the second about 2.5 minutes later, showing the publishing access was still live. Security firms Socket, StepSecurity and Ox Security documented the attack, and developers are advised to rebuild from the clean 1.26.0 release.
Read at BleepingComputer ↗