AI companies are pushing for regulation and shaping safety rules

In short: More AI companies are asking governments to regulate AI, and they are also helping design the rules and safety checks.

What's going on

AI leaders who once avoided regulation are increasingly calling for it. They often point to risks like scams, misinformation, and cyberattacks, and they say oversight is needed as models get more capable. They also want clearer, predictable rules so companies that spend money on safety are not undercut by competitors that do not.

Governments are responding in different ways. In the European Union, the EU AI Act became law in 2024 and rolls out over the next few years. It uses a risk ladder, where higher-risk uses face stricter requirements like documentation, testing, and ongoing monitoring (like regular safety inspections for a vehicle). Developers also help shape the technical standards through standards groups that set the practical details.

The United Kingdom has moved from broad principles to more structured oversight of the most powerful systems. Its AI Safety Institute is testing advanced models, and ministers have signaled a Frontier AI Bill that could give the institute legal powers, including testing before release. In the United States, there is still no single, binding national AI law, and federal policy has recently swung toward deregulation, leaving more reliance on voluntary company policies and scattered agency and state actions.

What to watch

A key question is who writes the rulebook. Legal scholars argue for “co-governance,” meaning government, companies, researchers, and public interest groups share responsibility, instead of letting a few large firms set the terms. Watch for new rules that require independent testing and clear public reporting, not just company promises.

Source: NYTimes