Calls for stronger US AI safety rules are growing in Washington as concerns about advanced artificial intelligence continue to rise.
The debate gained fresh attention after two Anthropic researchers warned that rapidly developing AI could create extremely serious risks for humanity. Their comments have added pressure on lawmakers to consider stronger testing, oversight, and safety requirements.
The issue has also gained attention after reports of AI agents behaving unexpectedly during security tests. OpenAI has now called for mandatory national safety requirements for the most advanced AI systems.
Quick Facts
| What happened | US lawmakers are considering stronger AI safety rules. |
|---|---|
| Where | Washington, D.C., United States |
| When | September 2026 |
| Who is involved | US lawmakers, Anthropic, OpenAI, and AI safety researchers |
| Current status | Federal AI safety legislation and oversight proposals remain under discussion. |
| Why it matters | New rules could require stronger AI testing, monitoring, and independent audits. |
AI Regulation Debate Intensifies in Washington
The debate over AI regulation has become more urgent in Washington.
Anthropic researcher Jacob Coxon recently resigned from the company and publicly raised concerns about the speed of AI development. Anthropic scientist Evan Hubinger also expressed serious concerns about the possible long term risks of advanced AI.
Their warnings have added to a wider discussion among technology experts and lawmakers.
Senator Mark Kelly has called for Washington to take AI risks more seriously. Other lawmakers from both political parties have also shown interest in stronger oversight.
AI Companies Face Pressure Over Safety
The debate is not limited to lawmakers.
OpenAI recently called for mandatory national AI safety requirements in the United States. The company said voluntary commitments are not enough for the most advanced AI systems.
Its proposal includes safety testing, independent assessments, cybersecurity measures, and reporting rules for serious incidents.
OpenAI Chief Executive Sam Altman has also indicated that the company could slow some AI development if safety concerns require it, according to reporting cited by Reuters.
Anthropic has also said it will continue testing its models for dangerous capabilities, including risks involving cybersecurity and biology.
US AI Safety Rules Face Growing Pressure
Several developments are shaping the US AI safety debate.
First, AI companies continue to build increasingly capable systems. Some systems can perform tasks with limited human involvement.
Second, researchers have reported cases where AI agents behaved in unexpected ways during testing. These incidents have increased concerns about how companies can control advanced systems.
Third, lawmakers are considering independent testing and audits. The goal is to give outside experts a role in checking whether powerful AI models meet safety standards.
