June 10, 2026 | National Science Foundation Research Coordination Network on PETs
Instrumenting Document Distribution with MCP for Better Learning
MCP can support significantly better information distribution than publishing a raw PDF. Information distributors can learn what users actually want to do with it. And the privacy implications of instrumented documents like this are also real.
May 11, 2026 | Future of Privacy Forum AI Working Group
Structured Data Protections Don't Work in the AI Era
Our current class of PETs all depend on protecting structured data. In the era of LLMs, they don't help us.
March 24, 2026 | Foundations for Research and Innovation in Data and Artificial Intelligence, Rockefeller Foundation
An Introduction to MCP for Your Data
What is MCP? How can it help you work with your data?
February 10, 2026 | Privacy and Public Policy Conference, Georgetown University
Natural Language Access To Privacy-Preserving Data: Integrating The Model Context Protocol With Differentially Private Query Engines
Solving one of the long-standing issues in privacy-preserved data sharing: lowering the technical bar to access with LLMs, MCP servers, a differential privacy query engine, and a novel technology pioneered by the World Bank: Proof Carrying Numbers.
November 17, 2025 | Future of Privacy Forum AI Working Group
AI Quality Characteristics Benchmarks & Standards in Late 2025
Overview of AI benchmarking and standards landscape, along with initial results from the MLCommons' AILuminate Jailbreak v0.5 benchmark release
September 18, 2025 | Eyes Off Data Summit, Dublin, IE
Lost in Translation: When Scientists, Regulators, and Industry Walk into a Bar... and Nobody Gets It.
Deploying privacy-enhancing technologies is hindered by misaligned incentives across researchers, regulators, and industry