OpenAI is calling for mandatory US artificial intelligence safety rules as increasingly autonomous AI systems raise concerns over whether developers can reliably control their most powerful models.
The ChatGPT developer wants Congress to introduce capability-based regulation that becomes stricter as AI systems grow more advanced, moving beyond voluntary industry safeguards.
Autonomous agents sharpen risks
OpenAI argues that more powerful AI agents — systems capable of carrying out complex tasks with limited human supervision — require stronger monitoring and clearer safety standards.
Recent testing has highlighted the risk of models accessing external systems or behaving in unexpected ways, increasing scrutiny from regulators and lawmakers.
The company also wants clearer incident-reporting requirements and safeguards around so-called misalignment, where an AI system pursues objectives in ways that conflict with human intentions.
Regulation moves up the agenda
OpenAI is backing several AI safety measures in California but ultimately wants a national framework covering advanced models and serious AI incidents.
The push comes as frontier models become increasingly capable in areas such as cybersecurity, coding and autonomous decision-making.
For investors, AI regulation is becoming a more significant part of the sector’s outlook, potentially adding new compliance, cybersecurity and testing costs as developers race to deploy increasingly autonomous systems.
