Posted 3/11/2024, 1:00:00 PM
Report Urges New Restrictions to Address AI Safety Risks
- Report commissioned by U.S. government says AI poses "extinction-level threat" and action must be taken quickly to avert risks
- Recommends new laws banning training of AI models above a certain level of computing power set by proposed federal AI agency
- Suggests outlawing publishing details of powerful AI models, with violations punishable by jail time
- Calls for tighter controls and monitoring of AI chip manufacturing and exports
- Report authors recognize recommendations are controversial but argue they are necessary given pace of AI progress