The Prof G Pod with Scott Galloway

Episode: Meredith Whittaker on Who Controls Your Data in the Age of AI
Date: March 5, 2026
Host: Scott Galloway
Guest: Meredith Whittaker, President of the Signal Foundation

Episode Overview

This episode features an in-depth conversation between Scott Galloway and Meredith Whittaker, president of the Signal Foundation and a prominent voice in tech and AI policy. The discussion explores the future of private communication, the risks of pervasive AI, the real meaning of encryption, the tension between privacy and public safety, and the evolving societal consequences of data collection. Whittaker provides insight on how technology companies shape our lives, who owns our data, and how we can regain agency.

Key Discussion Points & Insights

Signal’s Mission & How It Works

[04:13 – 06:22]

• Signal is designed from the ground up to be a truly private communications platform, collecting as little user data as possible—unlike most tech companies that monetize user information.
• Open source code provides transparency: “You don't have to trust me, you can actually verify...because we can scrutinize the code.” (Meredith, [06:04])
• Difference from WhatsApp: While WhatsApp uses the Signal protocol for message content, Signal goes further by encrypting all metadata (profile photos, contacts, who messages whom, group memberships)—not just message bodies.

  “WhatsApp only applies [encryption] to one layer…But WhatsApp does not encrypt intimate metadata…that’s very revealing data.” (Meredith, [07:20])

Encryption Misconceptions

[06:30 – 09:26]

• Not all “encrypted” apps provide the same level of protection. Many only partially encrypt data, especially metadata.
• Signal’s approach is all-encompassing—what Meredith refers to as “100% encrypted…not just sprinkling encryption dust on top of ultimately non-private infrastructure.” ([08:57])

AI, Agentic Risks & Privacy

[09:26 – 13:16]

• The integration of AI agents into operating systems (e.g., Microsoft Recall) introduces new privacy and security risks, undermining the guarantees that private platforms like Signal provide at the application layer.
• AI agents with broad device access can expose intimate data without needing to break encryption, creating significant vulnerabilities.

  “Instead of having to break our gold-standard encryption…you just have to leverage the type of access these agents are being given into your applications, into your intimate data…” (Meredith, [11:30])

• Because large models often require off-device processing, data frequently ends up on external servers, exposing it to companies or potentially bad actors.

Large Language Models & Data Risks

[17:17 – 18:49]

• Users should be cautious about what they enter into services like ChatGPT:

  “Any query to an LLM…is sending that data to servers controlled by OpenAI, Microsoft…They retain that data, they could leak that data.” (Meredith, [17:31])

• Legal and policy shifts could retroactively make previously innocuous queries incriminating.

"AI" as a Marketing Term

[18:49 – 21:55]

• AI is not a technical term of art, but a term with a history as “a kind of flashy term” for grant-getting and narrative building:

  “It allows us to step back and actually recognize that this is not a term of art…we can actually sort of have a bit more agency to define what we mean by intelligence.” (Meredith, [21:17])

The Real Threats of AI

[21:55 – 25:45]

• Most concern isn’t dystopian scenarios of robot overlords but the concentration of power:

  “Highly centralized technologies…rebranded as a kind of God’s head intelligence…making us less critical than we need to be.” (Meredith, [25:10])

• The real issue is using these systems in high-stakes, infrastructure-level decisions controlled by a handful of companies.

AI & The Labor Market

[25:45 – 30:11]

• AI used as a pretext for layoffs:

  “AI has been a handy pretext for job cuts... there’s some AI-wrapping of downsizing that is happening.” (Meredith, [26:35])

• Jobs may shift to lower-quality roles: Copywriters and translators become editors of AI output, reducing job satisfaction and security.
• Engineering concern: Outsourcing too much to AI agents may create unmanageable “technical debt” as only experts can maintain and understand the code.

Privacy vs. Public Safety & Government Surveillance

[33:05 – 37:31]

• Encryption either works for everyone or no one:

  “If you undermine the math of encryption…that’s breaking encryption for everyone…The people you hate the most have to be able to use it if the people you love the most are going to have access to it as well.” (Meredith, [34:51])

• The issue isn’t insufficient data for surveillance, but rather too much data ("finding the needle in the haystack"), with a shrinking refuge for private communication.

Policy & Regulation

[37:31 – 39:51]

• Meaningful consent should go beyond superficial agreement—companies shouldn't have the default right to create data about people at all.

  “The core issue is the authority we’ve given tech companies…to sort us and order us and tell us our place in the world.” (Meredith, [39:20])

Societal Attitudes Toward Privacy & Utility

[42:57 – 46:40]

• Apparent trade-off: People seem to surrender privacy for convenience, but Meredith reframes this as a lack of meaningful choice due to structural factors:

  “We use what we can to be together, to connect with each other, to participate in life. Those services have themselves…betrayed us structurally. And that doesn’t mean we don’t care about privacy." (Meredith, [45:14])

• Usage of Signal continues to rise, suggesting increased privacy awareness.

Notable Quotes & Memorable Moments

• On Encryption:
  “We have gone out of our way to be unalloyed, 100% encrypted… We're not just sprinkling encryption dust on top of ultimately non-private infrastructure.”
  —Meredith Whittaker [08:56]

• On AI Risk:
  “Our concern is really coming from a privacy integrity standpoint…these tools, which can be useful…but also pose this pretty significant risk that isn’t getting the kind of attention I believe it should.”
  —Meredith Whittaker [13:10]

• On Loss of Data Agency:
  “We have given tech companies who create data for advertisers the authority to sort and order our world and tell our stories for us.”
  —Meredith Whittaker [39:18]

• On Human Connection and Privacy:
  "Humans want to be loved and they want to be included….a meaningful choice around what it would take to care about privacy has not really been given to us.”
  —Meredith Whittaker [45:14]

• Galloway on Privacy Complacency:
  “I realize how promiscuous and careless I’ve been with my own data and I thought what I do is just not that interesting…most recently, when I hear the Trump administration talking about assembling lists…”
  —Scott Galloway [37:31]

• Closing Exchange:
  “Let’s put Meredith Whitaker in charge. Let’s just, let’s consolidate all of it. I’ll go raise $11 trillion…Deal?”
  —Scott Galloway [47:16]
  “It’s a deal, Scott. It’s a deal. Looking forward to working with you.”
  —Meredith Whittaker [47:39]

Timestamps for Key Segments

| Timestamp | Topic | |------------|----------------------------------------------------| | 04:13–06:22| How Signal works & why it’s different | | 06:30–09:26| Encryption myths and WhatsApp comparison | | 09:26–13:16| AI agent risks & privacy implications | | 17:17–18:49| Data exposure with LLMs like ChatGPT/Claude | | 18:49–21:55| “AI” as history & marketing, not technical clarity | | 21:55–25:45| Centralization of power as the big AI risk | | 25:45–30:11| AI’s labor impact: real and perceived | | 33:05–37:31| Privacy vs public safety; universal encryption | | 37:31–39:51| Desirable regulation: true consent; data creation | | 42:57–46:40| Privacy vs utility—a false choice | | 47:16–47:39| Closing: “Put Meredith in charge" banter |

Tone and Takeaways

• Whitaker’s voice is confident, nuanced, often philosophical, and grounded in technical expertise.
• Galloway is candid, humorous, and sometimes self-deprecating, pressing for clarity and practical guidance.
• Listeners leave with heightened awareness of privacy issues, skepticism toward tech company motives, and practical cautions about engagement with AI-powered products.

Recommended Action:
If you value privacy—or even if you think you have “nothing to hide”—use Signal, be cautious about the data you share with AI services, and push for meaningful choices in how your data is collected and used.

End of Summary.