Open-Weight AI Models Enable Low-Cost Cyberattacks, Experiment Shows

A US cybersecurity firm's experiment demonstrates that open-weight AI models can be used to conduct cyberattacks at a cost as low as three dollars, challenging the assumption that closed-source models are superior for offensive capabilities. The finding underscores the growing threat from accessible AI tools, though the specific attack types and model details remain undisclosed.

AI disclosure: Model-written summary of the source, not a verified fact. Check the sources. How Mosaic works