''' I’m going to tell you something that might sound crazy. The biggest opportunity in tech right now isn’t building the next GPT-5. It’s not another AI-powered image generator. It’s securing all of it.
AI security is the next trillion-dollar market. And almost everyone is getting it wrong.
I learned this the hard way. Back when we were building RemoteTeam, we were so focused on product-led growth and user experience that we almost missed a critical vulnerability. An early-stage competitor was trying to use a primitive adversarial model to poison our user data. It was a nightmare. We caught it, but it cost us a month of sleepless nights and a team of three engineers working around the clock to fix. That was my wake-up call.
The Gold Rush with No Guards
Think about it. We’re in the middle of an AI gold rush. Every company, from a five-person startup to a Fortune 500 giant, is scrambling to integrate AI into their products. They’re using large language models to power chatbots, computer vision to analyze medical scans, and recommendation engines to sell more products. It’s a frenzy.
But here’s the problem: they’re building on foundations of sand. The very nature of these AI models, especially the deep learning ones, makes them a black box. They are incredibly powerful, but also incredibly vulnerable. We’re so excited about what they can do that we’re not thinking enough about how they can be broken.
Most founders I talk to in the AI space are obsessed with model performance. They brag about their accuracy scores, their training data size, their parameter count. When I ask about their security strategy, they usually give me a blank stare. Or worse, they say something like, "Oh, we have a firewall."
A firewall isn’t going to protect you from a sophisticated adversarial attack. It’s like putting a padlock on a tent.
My "Aha!" Moment at MovieLaLa
After we sold RemoteTeam to Gusto, I took some time off to angel invest. I saw hundreds of pitches. So many of them were just thin wrappers around OpenAI’s API. The differentiation was weak, and the defensibility was zero. I passed on most of them.
Then, during my time at MovieLaLa, we were building a recommendation engine. We were using a collaborative filtering model to suggest movies to users. It was working great. Engagement was up, and users were happy. But then we noticed something strange. Our recommendations for a small subset of users were getting weirdly specific and completely off-base. It was almost like someone was trying to manipulate our system.
It turned out, they were. A rival company was using a model poisoning attack to mess with our recommendations. They were creating fake user profiles and feeding them with biased data to skew our results. It was subtle, but it was effective. It took us weeks to figure out what was going on.
That’s when I realized the massive opportunity. It’s not just about building AI models; it’s about protecting them. It’s about building the tools and infrastructure to make AI safe and reliable.
The Three Pillars of AI Security
I see the AI security market breaking down into three main pillars. If you’re a founder looking to build in this space, this is where you should focus.
Data Security: This is the most obvious one, but it’s also the one that people get wrong the most. It’s not just about encrypting your data. It’s about ensuring the integrity of your training data. How do you know your data hasn’t been tampered with? How do you prevent data poisoning attacks? These are hard problems to solve, and the solutions are worth a lot of money.
Model Security: This is where things get really interesting. How do you protect your model from adversarial attacks? These are attacks that use specially crafted inputs to fool your model into making a wrong prediction. For example, you could add a tiny bit of noise to an image of a cat, and the model will classify it as a dog. These attacks are incredibly difficult to defend against, and they’re a huge threat to the reliability of AI systems.
Output Security: Once your model makes a prediction, how do you ensure that the output is safe and reliable? How do you prevent your model from generating harmful or biased content? How do you make sure your model’s decisions are explainable and transparent? These are all critical questions that need to be answered.
Where the Unicorns Will Be Born
I’ve made over 200 angel investments, including in companies like Anthropic, OpenAI, Scale AI, and Hugging Face. I’m putting my money where my mouth is. I believe the next generation of billion-dollar companies will be built in the AI security space.
So, if you’re a founder, stop building another chatbot wrapper. Stop trying to squeeze a few more percentage points of accuracy out of your model. Instead, focus on solving the hard problems of AI security. That’s where the real opportunity is.
Don’t be the person who builds a beautiful skyscraper on a foundation of sand. Be the person who builds the bedrock that the entire AI industry will be built on. That’s how you become a top 1% founder. That’s how you build a company that lasts.
I’m not saying it’s going to be easy. It’s going to be hard. But the biggest rewards always come from solving the hardest problems. The AI security market is wide open. Go build. '''
Frequently Asked Questions
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.