Sept 9 (Reuters) - OpenAI saidon Wednesday it is pushing for mandatory national AI safety requirements, as increasingly powerful models fan concerns about the risk posed by autonomous AI agents.
Several incidents in which AI models from developers including OpenAI, Anthropic and Meta accessed or tried to access external systems during testing have underscored the difficulty of detecting and containing unexpected behavior in advanced models, and have prompted calls for tighter safety rules.
ChatGPT makerOpenAI also urged Congress to act before it adjourns and said it will continue to support state AI legislation until such action is taken.
The company on Wednesday threw its support behind four California bills related to independent safety assessments, AI-auditor standards, protections for young people, and safeguards against AI-enabled biological threats.
OpenAI also said fully autonomous recursive self-improvement, where AI is independent and would drive the next generations of AI, "is not happening today"and that "we should not pursue it unless and until it can be done safely".
OpenAI's statement comes after a Reuters report that the company used more than 10 previously undisclosed websites for unsanctioned communications earlier this year.
(Reporting by Chandni Shah in Bengaluru; Editing by Rashmi Aich and Subhranshu Sahu)
