Andrew Woods, GC of SSP Pubmatic, joins Alan Chapell to discuss Pubmatic’s approach to agentic ads — and how AI has irrevocably changed the legal profession. Pubmatic’s recent agentic test campaign results were a hit at Marketecture Live, and Andrew goes deep into what he believes will need to happen in order for agentic to live up to its promise.

The Chapell Regulatory Insider is available at https://chapellreport.substack.com/.

Takeaways

Agentic AI does not create new liability; it accelerates existing frameworks
Accountability sits with whoever deploys the agent
Logging and decision provenance will be critical for compliance
Privacy compliance is less about difficulty, more about fragmentation
Hope is not a plan when it comes to regulatory scrutiny
Data minimization is becoming a real engineering constraint
AI will eliminate routine legal work but elevate strategic problem-solving
Agentic systems can become powerful compliance tools if built correctly

Chapters

00:10 Introduction to the episode and Andrew Woods

02:01 Andrew’s background and journey into ad tech

04:20 Lessons from Twitter on privacy and user safety

06:29 What a General Counsel does in modern ad tech

08:43 Biggest lesson from global privacy enforcement

11:04 Building trust between legal and business teams

13:21 Who is liable in agentic AI advertising

16:29 Why logging and data provenance matter

18:27 The role of compliance signals and guardrails

20:57 Why agentic compliance frameworks are not built yet

22:57 Regulation versus fragmentation challenges

24:55 How PubMatic approaches privacy by design

27:10 Data minimization in practice

28:16 Non-negotiables for agentic compliance frameworks

31:57 Where industry discussions are happening

33:50 The future of agentic AI in advertising

36:42 How PubMatic’s legal team uses AI today

40:26 Why traditional legal work is disappearing

45:09 Final thoughts and closing

Learn more about your ad choices. Visit megaphone.fm/adchoices

Episode 73: Pubmatic isn’t messing around with Agentic Ads