Two former OpenAI researchers, Daniel Kokotajlo and William Saunders, who resigned this year over safety concerns are disappointed by OpenAI’s decision to oppose California’s bill to prevent AI disasters, SB 1047. They previously warned that OpenAI is in a “reckless” race for dominance and called for AI regulation. However, now that regulation is on the table, OpenAI has chosen to oppose it, which has left them disappointed.
In response to the former employees, an OpenAI spokesperson stated that the startup strongly disagrees with the mischaracterization of their position on SB 1047. They have endorsed AI bills in Congress and believe that frontier AI safety regulations should be implemented at the federal level due to national security and competitiveness implications. This response seems to highlight a disconnect between OpenAI’s public image and their actual stance on AI regulation.
Anthropic’s Support for the Bill
On the other hand, OpenAI rival Anthropic has expressed support for the bill while also presenting specific concerns and asking for amendments. Some of these concerns have been incorporated into the bill. Anthropic’s CEO Dario Amodei wrote to Governor Gavin Newsom, stating that the benefits of the current bill likely outweigh its costs, although not fully endorsing it. This shows a more nuanced approach to the issue of AI regulation compared to OpenAI’s opposition.
The disappointment expressed by the former researchers raises questions about whether OpenAI can truly live up to its mission statement of building AGI safely without adequate regulation in place. By opposing the bill aimed at preventing AI disasters, OpenAI may be sending conflicting messages about their commitment to AI safety and ethical development practices. This could potentially damage their reputation in the AI research community and beyond.
OpenAI’s decision to oppose California’s AI safety bill has been met with disappointment and criticism from former employees and industry rivals. The contrasting responses from OpenAI and Anthropic shed light on the complexities of regulating AI technology and the different approaches taken by organizations in the field. Moving forward, it is essential for OpenAI to reconsider its stance on AI regulation and work towards building trust and credibility in the AI research community.