China’s DeepSeek AI and the Future of AI Innovation & Regulation

The AI race just got a lot more interesting.

A Chinese startup, DeepSeek, has developed an AI model that rivals OpenAI’s ChatGPT and Google’s Gemini—but at a fraction of the cost. Unlike U.S. models that rely on expensive, high-performance Nvidia chips, DeepSeek had to innovate around U.S. export restrictions. The result? A more efficient AI model that costs just $5.6 million to train, compared to the hundreds of millions or even billions spent by leading U.S. AI firms.

That’s where my conversation with Inside the Issues host Amrit Singh began.

Necessity Breeds Innovation

U.S. export controls were meant to limit China’s access to advanced AI chips, but DeepSeek turned the challenge into an opportunity. They developed a model that operates with far less computing power—something that could disrupt the AI industry in unexpected ways.

And they didn’t stop there. Unlike OpenAI and Google, which tightly control access to their models, DeepSeek made its model open-source. That means developers can run it on local computers without relying on large-scale AI data centers. This could be a huge boost for AI innovation—but it also raises serious concerns.

Security & National Security Risks

DeepSeek's popularity is undeniable—it’s now the most downloaded app on Apple’s platform—but it also raises data security questions. If users rely on DeepSeek’s servers, their data is being processed in China. That’s a major risk. As I explained in the interview, there is a distinction between running the open-source model on your own infrastructure versus using DeepSeek apps and infrastructure. The broader implications of widespread Chinese AI adoption should not be ignored.

The U.S. government has already expressed national security concerns over Chinese-owned tech platforms like TikTok. With DeepSeek, similar issues arise:

• Data tracking risks – Could the Chinese government access and analyze user data from DeepSeek?

• Algorithmic manipulation – Will AI-generated content be influenced to favor Chinese narratives?

• Geopolitical leverage – Could this technology shift the balance of power in AI development?

The TikTok Precedent: When AI and Policy Collide

The discussion naturally turned to TikTok, which the Biden administration targeted with new regulations upheld by the Supreme Court. Historically, the U.S. has protected First Amendment rights, even when it meant allowing Americans access to foreign propaganda. But in TikTok’s case, the Supreme Court sided with national security concerns over free speech arguments.

As I pointed out, this decision sets a major precedent. The ruling affirms that the government can regulate foreign-owned tech platforms if national security is at stake. Could similar scrutiny soon apply to DeepSeek?

What Happens Next?

With TikTok, President Trump extended the deadline for compliance, but a shutdown still looms. DeepSeek may face similar regulatory hurdles, particularly if concerns grow over data security and foreign influence.

One option that’s been floated for TikTok is a forced divestment—essentially, the U.S. taking a 50% ownership stake. That’s a bizarre proposal, given the administration’s stance on privatization. Would the U.S. government really want to own and regulate a social media app?

My prediction? We’ll likely see a shutdown or significant restrictions rather than a regulatory compromise. China isn’t about to give up control of its social media darling, and the U.S. won’t risk its national security over an app.