Privacy engineering mid-year temperature check

Privacy engineering mid-year temperature check

https://iapp.org/news/a/privacy-engineering-mid-year-temperature-check

Publish Date: 2026-06-22 10:40:00

Source Domain: iapp.org

• Managing AI Privacy Risk

  • Increased focus on addressing privacy risks associated with AI.
  • Introduction of new practices like model cards and provenance standards like C2PA for AI models and outputs.
  • Adoption of cutting-edge technologies like differential privacy and secure compute environments to limit data usage and enhance data protection.

• Challenges

  • Complexity in tracking and handling AI training data.
  • Difficulty in using PETs such as differential privacy and secure compute methods due to high-dimensional data and complex scenarios.
  • Privacy vs. utility trade-offs remain unclear and persistent with lack of mature implementations.

• Using Large Language Models (LLMs) to Manage Privacy Risk

  • LLMs helping to identify and document privacy risks, serve as a supplemental aid in the traditional privacy risk assessment process.
  • More complex workflows are using LLMs to interpret privacy requirements and implement verifiable controls.

• Challenges

  • Expensive and time-consuming to make legacy systems AI-ready.
  • Risks of over-reliance on AI outputs and the potential for code bases and internal secrets exposure.

• Traditional Privacy Engineering

  • Ongoing activity in data hygiene, handling data flows, and managing data mappings.
  • Challenges in tracking data, managing consent, and ensuring data portability and usability.

• Looking Ahead

  • Opportunities for advancing age verification and technical controls within privacy engineering.
  • Encouraging more collaboration and knowledge-sharing to set new standards and drive improvements in privacy infrastructure.