Microsoft reports ARTEX used in financial attacks

As cyberattacks increasingly leverage open-source AI agent frameworks like ARTEX to execute multi-stage operations at machine speed, defenders are also turning to AI agents for detection and response. The author argues that effective security requires separating model reasoning from execution authority, combining behavioral detection with policy-driven authorization and verifiable automated workflows rather than relying solely on model capability or slow human intervention. Open article →