Skip to playerSkip to main content
  • 5 minutes ago
Nvidia and more than 20 technology companies launched an AI safety initiative focused on open-weight models. The move follows an OpenAI-linked cyberattack that exposed limits of closed systems.
Transcript
00:00It's Benzinga, bringing Wall Street to Main Street.
00:02NVIDIA and more than 20 technology companies launched a new artificial intelligence safety
00:07initiative focused on open-weight models after a cyber attack involving rogue open AI models
00:12exposed challenges with closed AI systems, according to CNBC. Hugging Face was unable
00:18to use leading US frontier models to defend against the attack because their guardrails
00:23did not distinguish between the attacker and defender. It instead used a self-hosted Chinese
00:28open-weight model that did not have the same restrictions. The initiative comes as US
00:33lawmakers consider restrictions on Chinese AI models and Treasury Secretary Scott Besant
00:38threatens sanctions on Chinese companies that carry out AI distillation attacks.
00:43NVIDIA said the incident showed defenders need the ability to inspect, adapt, and run advanced
00:48AI models on their own infrastructure. For all things money, visit Benzinga.com.
Comments

Recommended