Craig Newmark sees an AI bubble. It may not be a bust.

Summary

POLITICO Tech – Craig Newmark sees an AI bubble. It may not be a bust.
Date: December 4, 2025
Host: Stephen Overly
Guest: Craig Newmark (Founder of Craigslist, philanthropist)

---

Episode Overview

In this episode, host Stephen Overly sits down with Craig Newmark—Craigslist founder, philanthropist, and self-described nerd—on the sidelines of the Aspen Cyber Summit. Their wide-ranging conversation explores Newmark’s take on the current AI “bubble,” how emerging technologies are fueling new waves of fraud and disinformation, and the responsibilities of tech companies and philanthropists in shaping a safer and more trustworthy digital future.

---

Key Discussion Points & Insights

1. Are We in an AI Bubble?

• Parallels to Dot Com Era: Newmark draws a comparison between today’s AI investment frenzy and the dot com bubble, highlighting both hype and unresolved dangers.

  • Notable Quote (Craig Newmark, 02:05):
    “I think we are seeing a bubble here because there’s a lot of issues which haven’t been addressed, like hallucinations which could get people hurt, or not seeing the right kind of guardrails but preventing deepfakes of all sorts.”

• Unique Threats from AI: He emphasizes specific risks:

  • Hallucinations (AI-generated false information)
  • Deepfakes capable of enabling scams and cyber breaches

• Personal Precautions:

  • Newmark humorously shares:
    “I’m worried about this to the extent that Mrs. Newmark and I have a safe word and now and then I remind her of it.” (Craig Newmark, 02:23)

---

2. Cybersecurity Basics for the AI Era

• Newmark’s Advice:

  • Pause and reflect—a behavioral check before responding to online requests can prevent many scams.
    • “When you think you’re being approached, pause and take maybe nine seconds to reflect on what’s happening. There’s behavioral work which suggests that if you do that, you’ll avoid some scams…” (Craig Newmark, 03:15)
  • Use available security tools: single sign-on, passkeys, multi-factor authentication, system updates
  • Focus on broad education: Public must see themselves as part of the solution

• Early Internet Observations:

  • Newmark witnessed abuse and fraud since the Internet’s inception and believes bad actors, though few in number, are increasingly active and damaging.

---

3. The Tech Industry’s Responsibility

• Expectations for Tech Companies:

  • Providers “should step up, act in goodwill and do what they can to prevent scams.” (Craig Newmark, 05:16)
  • Support real-time detection, participate in knowledge sharing through groups like the Global Signal Exchange
  • Use AI and LLMs (large language models) to improve fraud detection, but do so carefully to avoid arming scammers.

• Challenge of Staying Ahead:

  • “The bad guys will see what you’re doing and then develop countermeasures. … there’s more to be done and you just got to keep acting in good faith, build up trust and safety and do what you can.” (Craig Newmark, 07:38)

---

4. Transparency vs. Security: Publishing AI Models

• Open Source Cautiously:

  • Newmark supports releasing AI models (including code, weights, biases) openly, but with careful controls to avoid misuse.
    • “I’d like to see them publishing in open source mode … but in a very careful way. … We want to prevent bad guys from getting that.” (Craig Newmark, 08:38)

• Efficiency Matters:

  • Tech note: “You don’t have to compute things in 32 bit precision. 8 bit suffices right now. Thank you for letting me feel technical, right?” (Craig Newmark, 09:17)

---

5. Business Models, Profit, and Safety

• Are Profit and Safety at Odds?
  • Overly questions whether tech’s focus on profit undermines product safety.
  • Newmark’s take: Companies must act in good conscience, sometimes at the expense of profit.
    • Legal liability and insurance pressure may force companies to act in good faith if voluntary action lags.
    • “Unless a company is acting in good faith, there may be liability issues. … Insurance companies will say coverage may only be there if you act in good faith to protect your community.” (Craig Newmark, 10:25)

---

6. Combating the Lucrative Fraud Economy

• Industry Scale:
  • Overly references a Reuters report on Meta allegedly earning $16B from fraudulent ads.
    • Newmark doubles down: Big platforms need better spam/scam detection, with LLMs as a vital tool.

---

7. Philanthropy, Networks, and Legacy

• Newmark’s Philanthropic Approach:

  • Leverages a "network of networks" in his giving strategy, relying on experts: “Frankly, as a philanthropist, I don't know what I'm doing. But with a network of networks of people who do.” (Craig Newmark, 12:56)
  • Embraces "beginner's mind": “I'm keeping a permanent beginner's mind because I do have networks who are getting the job done for me so I can sink quietly into my dotage.” (Craig Newmark, 13:16)

• Desired Legacy:

  • Wants the internet to be safe and useful for all, stressing the importance of detection and user alerts for suspicious behavior.
  • “With the cooperation of good actors in an industry, ... the Internet can be a relatively safe place to get things done.” (Craig Newmark, 13:50)

---

8. Optimism for the Internet’s Future

• Fundamental Optimism:

  • “Most people are just fine. They’re okay. They just want a level playing field, a safe one where we can all work together.” (Craig Newmark, 14:54)

• Is Philanthropy a Permanent Need?

  • Cites a conversation with Valas Dar (McGovern Foundation); Newmark agrees philanthropy should, ideally, become unnecessary, but doubts it will happen within his lifetime due to persistent societal challenges.

---

Notable Quotes & Memorable Moments

• “[AI] hallucinations could get people hurt, or not seeing the right kind of guardrails but preventing deepfakes of all sorts.” (Craig Newmark, 02:08)

• “The basics are things like, well, most fundamental. When you think you’re being approached, pause and take maybe nine seconds to reflect on what's happening.” (Craig Newmark, 03:15)

• “[Bad actors] are very loud and very active. ... People of goodwill need to work together to fight bad actors.” (Craig Newmark, 03:56)

• “A nerd's gotta do what a nerd's gotta do.” (Craig Newmark, 06:51)

• “You don't want to tell the bad actors what you're doing to stop them because they are smart and they are adaptable.” (Craig Newmark, 08:13)

• “Unless a company is acting in good faith, there may be liability issues.” (Craig Newmark, 10:28)

• “Frankly, as a philanthropist, I don't know what I'm doing. But with a network of networks of people who do.” (Craig Newmark, 12:56)

• “Most people are just fine … They just want a level playing field, a safe one where we can all work together.” (Craig Newmark, 14:54)

• Sign-off: “Like the Batman says, I’m not the nerd you want, but I’m the nerd you got.” (Craig Newmark, 16:24)

---

Key Timestamps

• 02:05 – Newmark on AI bubble and risks (hallucinations, deepfakes)
• 03:02–03:45 – Cybersecurity basics and behavioral advice
• 04:49–05:15 – Tech industry responsibility for scams
• 05:15–08:34 – How tech giants and social media companies could address fraud
• 08:38 – Publishing AI models and safe open source
• 10:04 – Business models vs. safety and legal/insurance impacts
• 12:23–12:48 – Scale of fraud and need for better detection (referencing Meta/Reuters)
• 12:56–13:39 – Philanthropic approach and “network of networks”
• 13:50–14:54 – Legacy, impact, and optimism for safe internet
• 15:41 – Should philanthropy be temporary?
• 16:24 – Memorable “nerd you got” Batman quote and closing

---

Summary

Craig Newmark remains a keen observer and advocate for digital safety, calling out the AI hype cycle while insisting on the internet’s potential for good if industry players act in genuine good faith. By encouraging basic digital hygiene, urging responsible technology deployment, and leveraging philanthropic networks for long-term impact, he presses for collaborative solutions against the dark side of rapid innovation. Yet, a note of optimism pervades—a belief that with collective vigilance, the internet can remain a platform where “most people are just fine”—safe, empowered, and connected.

Transcript

[A]

Meta is investing $600 billion in AI infrastructure, bringing jobs to communities across America. Phil a Meta building engineer in Las Lunas, New Mexico, says welcoming Meta into our community is creating more opportunities. Learn [e-mail] buildingamerica.

[B]

Hey, welcome back to Politico Tech. I'm your host, Stephen Overle, and on this show, I break down tech politics and policy with the people shaping our digital future. Craig Newmark is best known for shaping our digital past. He's the founder of Craigslist and a pioneer in online advertising. These days, he's giving away his vast wealth to organizations that fight disinformation and bolster cybersecurity. And he's especially keen on fighting online scams, an issue that plagued Craigslist in its early days and and since then has only gotten worse. On the show today, I sat down with Craig on the sidelines of the Aspen Cyber Summit to talk about fighting fraud, his philanthropic legacy, and how the rush toward AI today compares to the dot com boom and bust he saw 25 years ago. Here's our conversation.

Craig, welcome to Politico Tech.

[C]

Hey, I'm really glad to be here.

[B]

For anyone who may not know just your brief history, you started Craigslist, if I remember, in 1995, right, as an email list and then incorporated into a company in 1999. So you were very early on in not only Internet advertising, but the Internet period. You saw kind of the dot com bubble and bustle. We're in this moment now with AI where we see all this investment everyone's talking about the potential. Do you see kind of a new bubble forming like you saw in those early days?

[C]

I think we are seeing a bubble here because there's a lot of issues which haven't been addressed, like hallucinations which could get people hurt, or not seeing the right kind of guardrails, but preventing deepfakes of all sorts. And that's a pretty serious kind of thing because they are good enough to perpetuate scams or breaches in cyber defenses. I'm worried about this to the extent that Mrs. Newmark and I have a safe word and now and then I remind her of it.

[B]

Well, you. I know a lot of the work you're doing now on issues like fraud, disinformation, cybersecurity. I mean, these are all issues that you talk to experts. We talk to them on the podcast. They say it's only going to get worse because of AI. I wonder what is the AI future you see in that regard and through your work, through your philanthropy how are you trying to prepare for that?

[C]

Well, I've been talking to a lot of people, thinking through a lot of issues, and there's a lot of cybersecurity basics that everyone should be doing to keep themselves safe, their families safe, even their jobs safe, and even maybe help utilities, water and power stay safe. The basics are things like, well, most fundamental. When you think you're being approached, pause and take maybe nine seconds to reflect on what's happening. There's behavioral work which suggests that if you do that, you'll avoid some scams like you won't be scammed in the manner I was some years ago.

[B]

Is that how you got into this? Were you the victim yourself?

[C]

No, I saw it from the very beginning in the Internet because I've been working on the Internet pretty actively for over 30 years and I saw its potential in really big and good ways. But I also saw how people could abuse it. Now there are very few people comparatively who will abuse the net, but the people who will do so are very loud and very active. And people of goodwill need to work together to fight bad actors. The deal is that you do security basics again, pausing for a little bit. Then you got to do things like using the tools that are available to all of us, like single sign on pass keys, my multi factor authentication, update your systems. These are basics that everyone can do. We need to get the word out in a way that everyone can see they can be part of the solution.

[B]

So many of these scams and so much of this fraud that people encounter is perpetuated by social media or perpetuated online. And I'm curious, you know, as someone who has been, you know, involved in the Internet largely from the start of the consumer Internet, certainly what is the responsibility you see for tech companies today, for social media companies today to address these issues?

[C]

I think everyone providing services over the net should step up, act in goodwill and do what they can to prevent scams to stop them. And that means a lot of things like participating in efforts to detect signals that a scam might be happening and then work to prevent them in real time. For example, there are groups that publish the indications of possible scams like the Global Signal alliance, which is part of the, oh, part of the global anti scam alliance. So the Global Signal Exchange provides all sorts of these signals which companies could use to detect things which may be happening. I'm looking forward to AI being used to do this because there are companies that are detecting what are the tokens in the AI sense, which indicates suspicious behavior and Then they can tell somebody that something is a suspected scam. And in fact, I do get those messages, like today when my phone says, hey, something looks kind of sussier, as the kids say, and I should reconsider responding ever to that one. So the technology is emerging right now and I'm going to be digging pretty deep into this.

Because I can and because it's needed and, you know, because a nerd's gotta do what a nerd's gotta do well.

[B]

And you know, you're in, I think, kind of a unique position because I know in the very early days of Craigslist, right, you all had criticism as well about fraudulent posts or illegal sales and some of the things that we still see online today in a different format. And I wonder what you learned from that early experience that you think Internet companies today should be learning.

[C]

Well, the most fundamental lesson for everyone is that while things can be dangerous out there, most people are okay, they're good, they don't want to hurt you. But if you're providing services, I think you should exercise a good faith goodwill in stopping what you can. And the thing that I learned is there's always something you can do, then you can do better. The bad guys will see what you're doing and then develop countermeasures. And this involves cybersecurity and scam fighting. But the idea is that there's more to be done and you just got to keep acting in good faith, build up trust and safety and do what you can. I do think the large language model stuff does offer partial solutions in this regards. Partial solutions in the here and now. I'm just hoping that the companies doing AI who can get effective can publish their stuff to be shared, but in a controlled way, because you don't want to tell the bad actors what you're doing to stop them because they are smart and they are adaptable.

[B]

What's the solution you would like to see from them?

[C]

I'd like to see them publishing in open source mode. They're large language models including weights and biases and thresholds and the code to use that, but in a very careful way. Again, a lot of companies could implement that kind of stuff in real time.

But we want to prevent bad guys from getting that. And we want the implementation models to be ones which run economically without excessive power drain. And that involves the kind of stuff I learned in computer sciences 50 years ago, like you don't have to compute things in 32 bit precision. 8 bit suffices right now. Thank you for letting me feel technical, right?

[B]

I appreciate the technical notes. I'm not a technologist myself, but I do follow and I think it's a good point. And one question I come back to a lot is the business models in tech today are heavily centered on monetization and on using data and using AI to sort of extract as much profit as possible. Is that kind of economically incompatible with safety and with some of these sort of features we'd like to see from our technology today?

[C]

I think everyone has to act in good conscience and to protect their community, even if it's less profitable. It turns out some of the emerging technologies may help solve some of that. If that's not enough, the lawyers have figured out that unless a company is acting in good faith, there may be liability issues. I mean, I'm not a lawyer, but the lawyers are already launching lawsuits and that may be part of the solution. Or maybe the insurance companies will say coverage may only be there if you act in good faith to protect your community. Maybe that makes sense. This requires legal knowledge, which I lack. And again, I am not a lawyer, even though my high school tests said that maybe I should consider being one.

[B]

Not a computer scientist, but a lawyer would have been a very different career for you.

[C]

I don't think I'm tall enough to be a lawyer.

[B]

Foreign.

[A]

Infrastructure is bringing jobs to local communities like Las Lunas, New Mexico. Phil, who grew up in Las Lunas, has seen the positive impact that Meta's ongoing $600 billion investment in American jobs and infrastructure will bring. I had to travel for work, missing moments I can't get back. Then Meta opened a data center and brought new jobs. Now I don't worry about missing out anymore. Learn more about Meta's [e-mail] buildingamerica.

[B]

I was reading a recent report, you probably saw it too, from Reuters, where Meta had allegedly predicted it would make $16 billion from fraudulent ads and scams. My takeaway from that story was just, frankly, how big of a business it is for these bad actors to be online, to be conducting these kinds of scams. What's your sort of takeaway from that? How do we combat something that is so lucrative for bad actors, for tech companies, for all these players?

[C]

I think the big guys need to implement the spam detection methods that I was alluding to before. They already have some. They need to get much better. I think large language models trained properly really are part of that solution, and I think that will show the good faith required by whatever the lawyers will figure out.

[B]

What's your thesis or your ethos these days, when it comes to your philanthropic investments, what's most important to you?

[C]

What I've done is I've built a network of networks of people who are really good in my focus areas, and I rely on their advice to get things done. Frankly, as a philanthropist, I don't know what I'm doing. But with a network of networks of people who do.

Well, when a person is ignorant of something, if they get the help that they need, they can actually get something done. More charitably to myself, I guess that's beginner's mind. I'm keeping a permanent beginner's mind because I do have networks who are getting the job done for me so I can sink quietly into my dotage.

[B]

What would you like to see the impact be? I guess I imagine you provide some sort of guidance in terms of the legacy you want to have or the impact you want your giving to have.

[C]

Well, something part of the legacy of Craigslist is that it helps people get through the day, like putting food on the table. Craigslist showed a lot of people in our country that the Internet is actually useful and can be reasonably easy to use with my work now, I want to show people that with the cooperation of good actors in an industry, that the Internet can be a relatively safe place to get things done. I mean, if something happens on the street, it's going to happen online. But we can do better online because systems can detect signs of, as the kids say, sus behavior and then tell the user, the customer, hey, something looks bad here, maybe with an x percent probability, that's what I'd want to see, and then I can act accordingly.

[B]

Sounds like you're still very much an optimist than about the Internet and about technology in general. Is that fair to say?

[C]

I'm very much an optimist, if nothing else, that doing customer service for most of my adult life, like 50 years in the business, 40 years of customer service at ground level. I see that most people are just fine. They're okay. They just want a level playing field, a safe one where we can all work together.

[B]

I had an interesting conversation on the podcast a few weeks ago with Valas Dar, who's the president of the McGovern foundation, which does a lot of philanthropic giving. And he made the point that he would like to get to a place where philanthropy goes away because government and industry are sort of tackling some of these issues sufficiently on their own. Do you share that mentality?

[C]

Billis and the McGovern foundation do great work. I chat with him. I share that attitude. I just don't see it happening in my lifetime, so I just. I need to keep plugging away.

[B]

Why don't you see it happening? What do you think is the hurdle to get there?

[C]

There's a lot of challenges in the world these days, and I think philanthropy is going to play a role in a lot of these things, at least within the next 20 years. And I'm committed to doing this kind of thing, but only as long as I live. After that, it's over.

[B]

Well, listen, Craig, this has been a great conversation. I appreciate you being here on Politico Tech.

[C]

It's my pleasure. And like the Batman says, I'm not the nerd you want, but I'm the nerd you got.

[B]

The nerd we got. Thanks for being here.

That's all for this week's Politico Tech. If you like Politico Tech, go ahead and subscribe and recommend the show to a friend or colleague. And for more tech news, subscribe to our newsletters, Digital Future Daily and Morning Tech. Our producer is Normal Molaiko Cran Bandy made our theme music. I'm Stephen Overlee. See you back here next week.

Sam.