Skip to main content
Build trusted data with Ethyca.

Subject to Ethyca’s Privacy Policy, you agree to allow Ethyca to contact you via the email provided for scheduling and marketing purposes.

Issue 01

Hello World


Welcome to my new newsletter, Post Privacy. Those of you who know me will certainly understand that I do not believe that privacy is dead or unimportant. What is clear is that AI-driven business models, advanced data infrastructure, together with stepped up expectations from governments around the world, all come together to force organizations to broaden their horizons in how they govern data.

In this space I will focus on understanding three parallel trends:

  1. Changing User & Government Expectations: Growing demands from policymakers and enforcement authorities to take consumer rights seriously, especially requiring respect for the usage limitations and purpose specifications that underlie the user data contract. Enforcement authorities are no longer tolerating technical or operational gaps that leave consumers without control. Along with this, we’re seeing an expansion of policy attention from just a narrow class of privacy rights to broader fairness concerns raised by AI-driven systems.
  2. AI-Led Technical Change: Rapid changes in the business environment that push both technical and legal leaders to coordinate their efforts across the enterprise with more tightly integrated systems. End-to-end data governance is increasingly vital to realize the value locked inside enterprise data. Mobilizing data for AI enablement requires real-time decisions about whether data uses are permitted under the law and organizational rules. No matter how good privacy professionals and lawyers are, they cannot make individual decisions rapidly enough. Most enterprises are trying to find a way past two bad choices: either incur unbounded risk from data misuse or slow down AI deployment to unacceptably low levels.
  3. New Operating Models: Changing legal and technical requirements are forcing changes to firms’ operating models. How can the Chief Data Officer and Chief Information Officer, responsible for delivering agile, AI-driven engagement with the firm's data, work together efficiently with the General Counsel and Chief Privacy Officer to assure compliance at scale across the enterprise? The organizations I speak with in the market are all feeling pressure to adapt their current operating model for the times.

AI Applications Demand Better Purpose Specification

Even simple Agentic AI applications create a need for purpose specification and usage limitation enterprise-wide. Consider this very life-like use case: A sales analyst in the retail banking division launches an agent to learn which salespeople are the most highly paid. In the Direct Query (left-side) branch of this agentic workflow, the role-based access control deployed along with the Snowflake instance behind the MCP server prevents the agent from getting access to salary data because of the rule that only those in the HR department are allowed to see data.

Flow diagram of an agentic workflow

The agent adapts quickly so we see in the Clever Agent Workaround (right branch) the agent finds an email in the analyst's inbox that announces a sales team trip to Hawaii for the top 5% revenue producers. The date of the team trip is Dec 9-10, 2026. The agent is allowed access to inbox and calendar data across the entire sales division through the normal permissions the sales analyst already has, so it finds all those sales people who are in Hawaii on Dec 9-10 and produces a list of the top 5% revenue producers. The agent succeeds in circumventing the intended data policy even though the organization has role-based access control deployed.

Failure to enforce purpose specifications on data creates enforcement risk

This simple agentic AI example is case in point for the kind of pressure that our current data governance approaches face. Just in the last year, large enterprises have had to pay billions of dollars in fines and damages for failure to control personal data usage according to the purposes legally allowed under the terms of the original consent and governing privacy law. These companies all wish they had technical purposes specification and usage limitation enforcement in place:

  • General Motors was fined $220M by a group of State Attorneys General, $12.5M by California, and subject to strict line-of-business restrictions by the US Federal Trade Commission, all for misuse of driver location and telematics data. Data was collected for safety and accident prevention purposes but was misused for insurance underwriting and marketing, all in violation of the original purpose according to which it was collected.
  • Meta had to pay just the New Mexico Attorney General a fine of $25 Billion, $16 Billion of which was for misuse of data beyond the purpose for which it was collected.
  • Allstate Insurance and their Arity location data broker subsidiary are subject to a major investigation by the State of Texas, as well as several large class action law suits, most of which arise from the company’s failure to restrict driver data for the original purpose for which it was collected.

In all of these cases, better technical data governance controls would have enabled these companies to prevent the misuse of data beyond its authorized purpose.

AI Requires a New Operating Model for Data Governance

Alongside the need for scalable, agile data governance technology is a corresponding need for enterprises to re-think their operating models for data governance. Many large enterprises have highly professional Chief Privacy Officers and General Counsels, and technically outstanding data infrastructure operations led by CIOs or CDOs. However, there is often an organizational gap between these operating functions, with each engaged in governance activities as a kind of parallel play. Forward-looking enterprises are in the process of re-thinking their operating model to assure agile data governance.

We’re living through a period of exciting, if daunting, data governance challenges. In the early days of the commercial Internet and ecommerce, we saw the evolution of the Chief Privacy Officer, a role that was critical to help organizations navigate the legal, ethical and business challenges of expanded use of personal data. This vital role started with one leader: Harriet Pearson was appointed as IBM’s first Chief Privacy Officer in 2000. She was the first person in the world to hold this title. Chief Privacy Officers like Harriet were pathfinders in companies’ challenge to realize strategic value from personal data while respecting fundamental values and keeping their organizations out of legal hot water. Today, we are facing a similar set of challenges so I’ll be excited to share with you what I’m learning about how new leaders in technical, legal and business aspects of data governance are solving these problems anew.