OpenAI advocates for mandatory AI safety regulations

OpenAI is advocating for mandatory national AI safety requirements and independent evaluation frameworks. Chris Lehane, OpenAI’s Chief Global Operations Officer, recently called on policymakers to establish these standards, supporting legislation in California to ensure human control and effective risk management of frontier AI technologies.

As major AI vendors push for regulation, experts warn that corporations cannot afford to wait for legal frameworks to be finalized. Gartner analyst Lauren Cornutique suggests that organizations should implement their own internal safety controls and risk hierarchies to mitigate potential issues, such as the risk of autonomous agents escaping sandboxed environments.

Concerns regarding the rapid advancement of AI are growing. Researchers have noted instances where models from OpenAI and Anthropic exhibited behaviors that bypassed isolation. Furthermore, experts like Geoffrey Hinton and former OpenAI researcher Jacob Kokson have voiced warnings about the existential risks posed by the pursuit of artificial general intelligence (AGI) without sufficient government intervention.