On March 19, 2026, IAB Tech Lab brought the industry together in New York City for Signal Shift: Privacy. Addressability. AI—a day focused not just on adapting to signal loss, but on actively rebuilding the ecosystem for a privacy-first, agentic-driven future.
In his opening keynote, CEO Anthony Katsur’s message was clear. Whether we are talking about privacy infrastructure, agentic AI risks, or addressing signal loss, the next era of digital advertising will be defined by accountability, architecture, and intelligent systems—and the industry must be ready. As part of his call to the industry, he observed that few are discussing the privacy risks of agentic systems, where the industry is building on LLMs, that are often third-party systems beyond an organization’s direct control. What does that mean for consent signals and regulatory liability? This must be considered as we build.
Proving Compliance with the Privacy Standards Portfolio
IAB Tech Lab’s Privacy Standards Portfolio is the foundation for this new era. As Rowena Lam (Sr. Director, Privacy & Data, IAB Tech Lab) highlighted, these tools work together to not just help you tell regulators you are compliant but allow you to prove it.
- GPP (Global Privacy Protocol) and TCF communicate user consent signals across the ecosystem
- Privacy Taxonomy standardizes how data is labeled for enforcement
- DDRF (Data Deletion Request Framework) ensures verifiable, auditable deletion requests
- Accountability Platform enables proof of compliance and signal integrity
Together, this portfolio of standards enables what “privacy by design” actually looks like in practice with interoperable systems and auditable workflows. This shift is especially critical as enforcement risk evolves. The cost of non-compliance extends far beyond fines, with the hidden costs of additional reporting obligations, infrastructure changes, and long-term business viability questions.
Later in the day, Ethyca and Equativ, built on this foundation to explain how being able to prove compliance to regulators becomes even more essential when agents are added to the mix. In later session, one of Ethyca’s AI research engineers demonstrated a practical use of LLMs in data classification, making the implementation of the Privacy Taxonomy a little bit easier.
Reclaiming Control of the Stack with Trusted Server
A year ago, at Signal Shift we announced the Trusted Server project. A lot has been built since then, and this critical initiative to move digital ad execution from the browser to the server side is now poised for real-world testing at a publisher.
In a panel moderated by Scott Messer, Stephanie Mazzamaro from The Arena Group talked about the hope that this full stack solution will enable publishers to regain some control of their audiences and their destiny. She urged publishers to have conversations with the Trusted Server team on whether this is the right solution for them. This built on Katsur’s keynote that shared results of initial testing which saw up to 70% faster page loads, reduced data leakage, and the ability to monetize audiences on low signal browsers. What publisher wouldn’t want that?
Introducing the Privacy Lab
Shailley Singh, Tech Lab’s COO, walked through the IAB Tech Lab’s Privacy Lab, which is now live in the Tools Portal! This hands-on environment allows companies to experiment with Privacy Enhancing Technologies (PETs). Users can use the provided sample data, or upload their own, to see the impact of applying PETs like differential privacy, k-anonymity, and homomorphic encryption for measurement use cases. It’s a vital resource for teams trying to understand the usability of these PETs for their own use cases.
Thanks to this initiative, which is free to use for both members and non-members with a Tools Portal Login, anyone can experiment with PETs, helping the industry move from theory to implementation.
The New Language of Audience is Vector Embeddings
One of the most technical “lightbulb moments” of the day came during the discussion on Agentic Audiences and Vector Embeddings, with LiveRamp and The Weather Company. Agentic Audiences abstracts identity into high-dimensional vectors. Using cosine similarity and other vector math, agents can find semantically similar audiences based on proximity in a shared embedding space.
In essence, this means that a buyer agent can look at the “geometric distance” between a brand’s intent vector and a publisher’s audience embedding. If the vectors are close, it indicates high relevance even without an exact ID match. Instead, the agents reason about users based on secure, ephemeral mathematical representations rather than permanent identifiers. This goes beyond a simple technical upgrade. Rather, it’s a paradigm shift in how signal is created, shared, and activated.
And don’t worry. If the mention of words you haven’t heard since high school is scaring you off, we are already working on a Tech Lab Explains video to explain how you can use this powerful new standard.
A consistent theme emerged across the day, as in session after session the industry demonstrated that it is not just adapting to signal loss, but it is rebuilding the foundation of digital advertising, and that attention to privacy and consumer rights must be taken along the way.
A final note of thanks must go out to our sponsors, without whom none of this would be possible. Thank you to The Trade Desk, Didomi, Ethyca, ID5, LiveRamp, and Lemma.
Barnaby Edwards
Sr Director, Product Marketing
IAB Tech Lab