OpenAI has called for stronger global standards to manage the risks associated with increasingly autonomous artificial intelligence, including the possibility of recursive self-improvement (RSI).

In a recent post, OpenAI said fully autonomous RSI is not happening today and should not be pursued unless it can be developed safely while preserving meaningful human control. The company said AI is already accelerating parts of the research used to develop and align future models.

OpenAI has urged governments to develop common methods for measuring AI capabilities, evaluating safety risks and determining when human review should be required. It also proposed shared protocols for classifying and reporting serious AI incidents.

The company is calling for the United States to take a leading role in developing compatible international technical standards, while working with AI safety organisations and governments in other countries.

OpenAI also supports mandatory national AI safety requirements for frontier systems, including testing, independent assessments, cybersecurity measures and incident reporting. It argues that technical safeguards inside individual AI companies alone will not be sufficient as advanced systems become more capable and globally accessible.

The proposal comes as AI companies increasingly explore systems capable of conducting research and other complex tasks with greater autonomy. OpenAI says maintaining human oversight and establishing common safety standards will be important as this development continues.