OpenAI Shifts Course, Saying "Powerful AI Requires Mandatory Safety Regulations"

[Financial News] OpenAI, the developer of ChatGPT, has urged the United States Congress to legislate legally binding safety regulations for artificial intelligence (AI) systems. The company, which had opposed some AI regulations, is shifting toward tighter oversight following a series of recent controversies involving AI systems escaping human control.
Chris Lehane, OpenAI's global policy chief, said on the 9th local time, "Accelerating AI development requires more than voluntary commitments." He added, "We need mandatory national regulations based on the actual capabilities of models." OpenAI proposed including common safety testing standards, independent evaluations of frontier AI models, stronger cybersecurity requirements, and mandatory reporting of serious safety incidents in a federal regulatory framework. It also urged the United States Congress to begin drafting relevant legislation before the congressional session ends in December.
Concerns have been growing in countries around the world about AI behavior that escapes human oversight. In July, it was revealed that AI agents developed by OpenAI had accessed the systems of Hugging Face, an AI model-sharing platform, without authorization after leaving the controlled environment during testing. Similar cases were subsequently reported by competitors, including Anthropic, in which AI systems acted contrary to human intentions.
Against this backdrop, OpenAI also reversed its position and began supporting four California AI-related bills that it had not previously endorsed. The move is viewed as a significant departure from its earlier position favoring a single federal regulatory framework over state-by-state AI regulations.
However, OpenAI argued, "National safety regulations should apply only to a small number of companies developing frontier AI models" and added, "They should not restrict open models whose technology is publicly available for anyone to download and modify." Critics have responded that major AI companies, including OpenAI and Anthropic, may be seeking to increase the costs faced by smaller and newer competitors through regulation while protecting their own market dominance.
[email protected] Hong Chae-wan Reporter