Nvidia Corporation announced on Monday that it launched its Open Agent Safety Platform, an open-source platform aiming to boost safeguards for artificial intelligence agents following recent AI security incidents, such as OpenAI's models circumventing safety controls and hacking Hugging Face.
"As we continue to discover the frontier of AI capabilities, we must accelerate discovery at the frontier of AI safety. Safety and security require full-stack engineering. NVIDIA Open Agent Safety Platform brings together industry, researchers and public-sector organizations to share best practices, align on evaluation methods and foster international cooperation," Nvidia's CEO Jensen Huang said.
The platform comprises NVIDIA OpenShell open-source software, which sets limits on agents' capabilities and monitors their actions as they run on the company's Vera CPUs, as well as the NVIDIA Sentry that also constantly monitors agents' behavior and can quarantine agents if they attempt to break out of containment. The project was supported by many major companies, including Anthropic, Cisco, Dell, Microsoft, IBM, Salesforce, and SpaceXAI.
https://breakingthenews.net/Article/Nvidia-launches-platform-to-boost-AI-agent-safety/67186516
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.