# Can AI Be Regulated?

> As AI companies call for oversight, experts debate whether existing law, market forces, or government capacity can actually manage the technology's risks.
- Title: Can AI Be Regulated? · Foreign Policy | Shortify
- Summary: As AI companies call for oversight, experts debate whether existing law, market forces, or government capacity can actually manage the technology's risks. AI…
- Keywords: artificial intelligence, ai regulation, united states, openai, anthropic, nvidia, meta, donald trump, liability law, market forces, regulatory capture, government expertise
- Source: Foreign Policy — https://foreignpolicy.com/2026/09/24/can-ai-be-regulated
- Author: Cameron Abadi, Adam Tooze
- Read time: 3 min
- Topics: artificial intelligence, ai regulation, united states, openai, anthropic, nvidia, meta, donald trump, liability law, market forces, regulatory capture, government expertise
## Pressure builds for AI rules
The U.S. public increasingly wants government regulation of artificial intelligence, but President Trump opposes new restrictions. The debate centers on whether existing law can handle AI's risks.
## Negligence law might work
Existing legal frameworks like negligence, product liability, and consumer protection could regulate AI by holding companies accountable for releasing unpredictable systems with serious consequences.
## But AI breaks traditional law
AI algorithms are non-deterministic, meaning identical inputs don't always produce identical outputs. This makes it hard to prove causation and negligence in court.
## Extinction risks defy compensation
If AI causes civilization-ending harm, financial liability won't help. Courts may not exist to hear cases, and no amount of money can undo extinction-level events.
## Reporting gaps block enforcement
AI labs voluntarily report safety incidents, but creating legal liability could discourage honest disclosure. Without mandatory reporting, regulators can't know what's actually happening.
## Tech leaders reject regulation
Jensen Huang of Nvidia and Mark Zuckerberg of Meta argue that market forces already prevent unsafe AI releases. Companies won't launch products they don't trust.

> the market forces are already there, we don't need any new laws, we don't need new regulations
## The valuation paradox
AI companies are worth hundreds of billions because investors believe the technology will transform the world. But if AI turns out to be overhyped, the entire sector could collapse.
## Massive investment at stake
AI spending now drives 36 percent of all U.S. business investment, running at 150 to 200 billion dollars per quarter. A market reset would destabilize the entire economy.
## Big labs call for regulation
OpenAI and Anthropic are pushing for government oversight, which raises a suspicious question: are they trying to use regulation to crush smaller competitors?
## Regulatory capture theory
If regulation imposes high compliance costs, it protects dominant firms with sunk costs while blocking new entrants. This is a classic oligopoly strategy.
## Closed vs. open models
OpenAI and Anthropic use closed, proprietary models. Nvidia and Meta favor open models because open AI drives demand for computing chips and spreads AI globally.
## Chinese competition looms
Chinese AI labs produce 95 percent of the performance of American frontier models at a fraction of the cost. This threatens OpenAI and Anthropic's business model.
## The real regulatory question
The key issue isn't whether to regulate, but what kind of regulation. Will OpenAI and Anthropic allow external inspectors to examine their closed models?
## U.S. expertise is scattered
The government has pockets of AI expertise at NIST, national labs, the NSA, and DARPA, but they're fragmented across agencies and can't match private sector salaries.
## The salary problem
AI labs pay technicians millions of dollars. The government can't compete, so it loses talent to the private sector, just like in financial regulation.
## Asymmetric information
Regulators would need to understand machine learning, hardware, cybersecurity, biology, competition policy, and national security. AI labs know far more than government ever will.
## Trump complicates everything
The Trump administration's unclear AI strategy makes it harder to assess whether the U.S. government can even formulate coherent regulation.
## No clear answers yet
Existing law may be inadequate, market forces may be insufficient, and government capacity is questionable. The regulation debate remains genuinely unresolved.
## Key takeaway

AI regulation faces a fundamental tension: existing legal tools may be inadequate for catastrophic risks, while the companies pushing for rules may be using regulation to entrench their market dominance.