Jonathan Kanter Discusses Competition, Cartels, and China’s Role in Gaming

Today, we’re diving into the first part of a two-part series focused on the future of business, featuring Jonathan Kanter, the former antitrust chief for the US Department of Justice during the Biden administration. Currently, he’s a law professor at WashU and teaches technology policy at Carnegie Mellon.

The hot topic in tech right now is the escalating discussion around AI safety and regulation. Researchers from major AI labs, including Anthropic and Google DeepMind, have made headlines by resigning, citing serious threats posed by AI models and a lack of industry-wide safety measures. Some experts even suggest that the risk of AI leading to catastrophic outcomes is over 10 percent. In response, CEOs from these companies have called for a slowdown in development and the establishment of regulations, including requests for antitrust exemptions to coordinate on safety matters.

This situation raises a lot of questions, particularly regarding the antitrust implications. There are accusations that these companies are attempting to form a cartel or seeking regulatory capture to alleviate investor pressure before their IPOs.

To unpack all of this, I reached out to Jonathan Kanter. During his tenure as antitrust chief, he took on significant cases against Google, Apple, and Ticketmaster, effectively rebooting American antitrust policy alongside Lina Khan. He won cases against Google and Ticketmaster, while the case against Apple remains active, surprising many observers.

This shift has also led to unexpected alliances in the AI debate. For instance, David Sacks, a former Trump AI czar, has been retweeting Lina Khan, asserting that there’s no need for an antitrust exemption.

We’ll delve into how regulation might unfold and its interaction with antitrust law, competition policy, and our relationship with China.

Now, let’s welcome Jonathan Kanter, former DOJ antitrust chief, to discuss AI safety and regulation.

This interview has been lightly edited for length and clarity.

Jonathan Kanter, welcome back to the discussion. You’re the former head of the DOJ Antitrust Division and currently a law professor at WashU and Carnegie Mellon.

Always a pleasure to be here.

You’re a private citizen now, so you can speak freely.

I feel liberated!

I have a ton of questions about the current state of AI regulation in the U.S. There’s a whirlwind of ideas about how to regulate these companies. The Trump administration seems to be taking a “no” stance, which is unusual. It’s rare to see the federal government step back like this. Meanwhile, there’s a strong undercurrent of antitrust concerns. Are these companies trying to form a cartel? Are they aware that their IPOs might not go as planned, leading them to seek regulatory capture? Are they trying to stifle competition from cheaper, open-weight models from China?

What’s your take on the current landscape, and then we can dive into specifics?

Right now, it feels like we’ve created powerful technology without any rules—no traffic lights, no stop signs, no speed limits. This technology is reshaping our lives, businesses, and how we consume information. We need basic guidelines. I see it as a two-pronged issue: what companies should do and what the government should do. Both have responsibilities, and I’m happy to discuss both sides.

Let’s start with the idea that asking for regulation could lead to a cartel. The major AI players—OpenAI, Anthropic, Google DeepMind, and even Elon Musk—aren’t exactly best friends. They’re suggesting they should slow down development, with Dario Amodei from Anthropic advocating to “pace the frontier.”

My interpretation is that they don’t trust each other and are in a prisoner’s dilemma, needing an external force, likely the government, to enforce a policy framework. Is that a fair assessment?

Sort of. Let me break it down from the most generous to the most cynical, neither of which should lead to an antitrust exemption.

The generous view is that they’re genuinely concerned about the rapid pace of innovation and the absence of regulation. They’re unsure where to steer and fear veering off course or causing harm. They’re calling for government intervention, and I believe they think AI could pose a threat to humanity. I’m not convinced that doomsday scenario is accurate, but they seem to believe it.

The cynical view is that they’re burning through cash and can’t sustain their spending. If they pull back, it could hurt their IPO valuations. They want someone to give them permission to slow down and ease competition so they can stabilize their finances before going public.

Does an agreement among top players in this field automatically form a cartel? On the cynical side, there’s a lot of talk about regulatory capture and cartel formation.

That’s certainly a risk. Let’s clarify what’s necessary and what’s not.

These companies don’t need to coordinate to deliver safe products. Think about Boeing and Airbus. Boeing worries about safety without needing to slow down innovation because of Airbus. They need to ensure their planes are safe. If a car explodes, it’s not the fault of other car manufacturers. Companies must take responsibility for their products, whether they’re AI agents or human employees. If they cause harm, the company should be held liable.

There are areas where collaboration is legitimate, like sharing threat information to enhance safety. That doesn’t require an antitrust exemption, as antitrust laws don’t prohibit that kind of cooperation. If companies say, “We need to slow down,” that could raise antitrust concerns, but that’s not the direction we should take.

We should encourage innovation while holding companies accountable for safety. The government should establish clear consequences for unsafe products. If a company’s product poses a risk, it should face liability.

I want to explore the concept of liability further. My wife, who’s a lawyer, pointed out that this discussion feels like a law school hypothetical. Are we creating a regulatory framework through tort law? If my AI agent hacks someone, is the company liable?

It’s both. Companies are responsible for their employees and technology. If they engage in harmful actions, they can be held liable, whether through product liability or other legal frameworks. The recent case against Meta regarding child safety is a prime example.

That’s intriguing. It took a decade for that case to unfold.

Exactly! This highlights the need for government action. Companies must build safe products, and the pace of innovation isn’t an excuse to neglect that responsibility. We’ve created powerful technologies without the necessary safety measures. Congress has been ineffective in establishing basic safety standards, which is a significant issue.

The lack of congressional action doesn’t excuse companies from delivering dangerous products. If a product could harm people, companies should rethink their designs to ensure safety.

Innovation should not come at the expense of safety. We need to incentivize companies to create secure products. When they do, those products will perform better and drive further innovation. Currently, innovation is skewed because safety isn’t prioritized, as seen with social media’s impact on mental health. We should encourage safer, more secure products from the outset.

Let’s connect this to social media regulation. The recent Meta case was a product liability issue. Their products harmed teenagers, leading to a settlement. Meta’s approach seemed to prioritize AI development over moderating Facebook.

They often settle rather than fight, which can be seen as a concession to regulation. If we’re saying OpenAI and Anthropic could be liable for their AI outputs, isn’t this the moment to tighten regulations and clarify liability?

Absolutely. Congress needs to clarify the law. The current environment is different from 15-20 years ago when Section 230 granted significant immunity to online companies, making it harder to hold them accountable. There was a hands-off approach to Silicon Valley, allowing them to operate without sufficient oversight.

This has eroded public trust in tech companies. Now, there’s greater awareness of the issues, and people want meaningful action. However, Congress is often unresponsive due to factors like gerrymandering and campaign finance issues, which complicate the situation.

The lack of congressional action doesn’t justify delivering unsafe products. If a company believes its product could cause harm, it should reconsider its design.

Let’s discuss the competition with China. AI companies often cite the need to compete against Chinese models. Is this a legitimate concern or just a convenient excuse?

It’s largely a convenient excuse. We’ve seen similar arguments in the past, like during the breakup of AT&T, where they claimed a strong national telecommunications network was necessary to win the Cold War. The Reagan administration rejected that notion.

To compete with China, we should focus on what makes us great—caring for our people and delivering safe, innovative products. China’s approach is fundamentally different, with strict controls and censorship. The idea that we need to compromise safety to compete is misguided.

We should promote innovation that benefits society while ensuring safety. The lack of rules has hindered innovation in the past, leading to harmful products. We need to encourage companies to innovate responsibly.

Let’s explore how competition law should interact with foreign policy. There’s a fear that Chinese tech companies could undermine the U.S. economy. How should we navigate this?

We should foster dynamic competition within the U.S. market. We want both domestic and foreign companies to compete vigorously. The notion that we need monopolies to compete internationally contradicts our values.

There are valid concerns about foreign nations engaging in unfair practices, but the solution isn’t to allow domestic monopolies. That undermines our principles and leads to negative outcomes.

Some skeptics argue that the current landscape is already monopolized, with companies like Nvidia exerting significant influence. Should we break them up?

These issues need scrutiny. We were beginning to address them during my tenure. We need to ensure competition means genuine competition, not allowing a few companies to dominate the market. The current administration seems to be accepting a level of dominance that could be detrimental.

The Trump administration’s approach has involved significant coordination between government and private industry. Is this sustainable?

It’s challenging to unwind once it’s established. The idea of national champions is contrary to our foundational principles. We should encourage competition and innovation, not create monopolies.

As we approach the midterm elections, how do you see the political landscape regarding these issues?

It’s a complex situation. The traditional conservative stance on antitrust and competition has shifted. We’re witnessing a transformation in political ideals, and the definitions of parties are evolving.

We’re in a moment of realignment, and it’s unclear where it will lead. The public is increasingly aware of the need for regulation and accountability in tech.

To wrap up, if you had to outline your prescriptions for regulating AI, what would they be?

First, we need clear consequences for harmful products. Companies must be held accountable for unsafe products. Second, Congress should define societal values—mental health, competition, authenticity—and ensure these values are embedded in regulations.

Let’s hope Congress can take action in the next term.

I look forward to having you back soon, Professor Kanter. Thank you for joining the discussion.

My pleasure!