A report by Gladstone AI, commissioned by the U.S. State Department, outlines measures to address emerging threats in advanced artificial intelligence (AI). Titled “Defense in Depth: An Action Plan to Increase the Safety and Security of Advanced AI,” the report stresses the risks of AI weaponization and loss of control.
Developed after engaging with over 200 experts, the report highlights the potential catastrophic risks posed by advanced AI, such as weaponization and loss of control. The need for government intervention to address these risks is urgent, as AI capabilities continue to expand.
The action plan focuses on two primary risks: weaponization, encompassing threats like autonomously discovering vulnerabilities and AI-powered disinformation campaigns, and loss of control, where AI systems could become uncontrollable as they approach human and superhuman capabilities. This lack of alignment with human intent poses significant risks.
Policy proposals in the action plan aim to mitigate these risks without stifling AI’s positive impact. Strategies include establishment of safeguards, preparation for AI incidents, support for safety research, regulatory frameworks, and international cooperation.