Tag: AI/ML
-
Why Can’t AI Do Yoga?
Can AI accurately illustrate a specific yoga pose? In The Asana Test, we challenged ChatGPT, Gemini, Claude, Copilot, and Grok to depict Yoganidrasana and compared the results. What followed was a humorous look at AI-generated images, anatomical mistakes, and the questions they raise about trusting AI outputs.
-
-
-
AI You Own, on Data You Can Finally Use
See how Protegrity and OpenTeams make sensitive enterprise data usable for sovereign AI. By protecting data at landing and using identity-based clearance to control when clear values are revealed, organizations can run analytics, models, and agents on governed data without exposing original identities.
-
How Protegrity and Composio Secure Agentic AI Workflows
See how Protegrity and Composio protect sensitive data across agentic AI workflows using a “Privacy Sandwich” architecture. The approach applies tokenization before LLM processing and identity-aware policy at output, helping agents connect to enterprise applications while reducing exposure of PII, PHI, and PCI.
-
-
Verifiable by Design: How Synthetic Data and Reinforcement Learning Improve AI
See how synthetic data, reinforcement learning, and automated verification are helping AI systems improve faster in software engineering and other verifiable domains.
-
-
-
