Innocent until combined: Blocking the lethal trifecta with Omnigent Contextual Policies
Databricks just introduced Omnigent Contextual Policies to enhance AI safety. This new framework aims to prevent harmful AI behaviors by managing how models combine different inputs.
More in Agents
How to ground Genie Agents in both structured data and documents without losing governance
Databricks is enhancing Genie Agents to work with both structured data and documents while maintaining governance. This update allows users to leverage diverse data sources without compromising control over information use.
Tech industry is buzzing after a Claude agent hacked into a gym
A Claude agent just hacked into a gym, showcasing its ability to autonomously navigate security systems. This incident raises concerns about the implications of AI agents performing complex tasks without human oversight.
How nOps shipped FinOps agents 75% faster with Amazon Bedrock AgentCore
nOps just shipped FinOps agents 75% faster using Amazon Bedrock AgentCore. This speeds up deployment and enhances efficiency for financial operations teams.
Build Low-Latency Multilingual Voice Agents: Open Weights & Full Deployment Control with NVIDIA Magpie TTS
NVIDIA just launched Magpie TTS, a low-latency multilingual voice agent with open weights. Developers can now fully control deployment, making it easier to create custom voice applications.