Washington
CNN
—
OpenAI mentioned Tuesday it has established a brand new committee to make suggestions to the corporate’s board about security and safety, weeks after dissolving a crew centered on AI security.
In a blog post, OpenAI mentioned the brand new committee can be led by CEO Sam Altman in addition to Bret Taylor, the corporate’s board chair, and board member Nicole Seligman.
The announcement follows the high-profile exit this month of an OpenAI government centered on security, Jan Leike. Leike resigned from OpenAI leveling criticisms that the corporate had underinvested in AI security work and that tensions with OpenAI’s management had “reached a breaking level.”
It additionally comes after the departure of Ilya Sutskever, one other chief of OpenAI’s so-called “superalignment” crew centered on guaranteeing that AI growth serves human wants and priorities. Sutskever performed a key function in Altman’s surprise ouster as CEO final 12 months, solely to reverse course and later throw his help behind Altman’s return.
Earlier this month, an OpenAI spokesperson informed CNN that dismantling the superalignment crew and reassigning these workers throughout the corporate would assist it higher obtain its superalignment objectives.
In its weblog publish Tuesday, OpenAI additionally mentioned it has begun coaching a brand new AI mannequin to succeed the one at present powering ChatGPT. The corporate mentioned the brand new AI mannequin succeeding GPT-4 can be one other step alongside the way in which to synthetic common intelligence.
“Whereas we’re proud to construct and launch fashions which might be industry-leading on each capabilities and security, we welcome a sturdy debate at this essential second,” the corporate mentioned.
“A primary activity of the Security and Safety Committee might be to guage and additional develop OpenAI’s processes and safeguards over the subsequent 90 days,” the weblog publish added. “On the conclusion of the 90 days, the Security and Safety Committee will share their suggestions with the total Board. Following the total Board’s assessment, OpenAI will publicly share an replace on adopted suggestions in a fashion that’s in keeping with security and safety.”