OpenAI, the creator of ChatGPT and a leader in the Artificial Intelligence sector, reportedly rushed its new testing protocol for GPT4-Omni according to anonymous employees familiar with the process.
A representative of OpenAI’s preparedness team told the Washington Post the evaluations took place during the course of a single week. Although this was sufficient time to complete the tests, some employees felt as though the safety protocol was seen as an afterthought.
“They planned the launch after-party prior to knowing if it was safe to launch,” one employee told the WaPo.
These complaints over security are not new to emerging tech giant. OpenAI executive Jan Leike and research engineer William Saunders both resigned this year and have made public statements about the company’s prioritization of commercial goals over safety concerns. Leike, who exited the company after the release of GPT-40 posted to X, “Safety culture and processes have taken a backseat to shiny products.”
President Joe Biden signed an executive order last October to guide responsible AI development and deployment across the United States. Prior to this OpenAIwas one of a dozen tech companies that made voluntary commitments to the White House over AI safety.
An OpenAI spokesperson said that while the company did not “skirt” on safety protocol, the testing process was intense and would not happen again in such a tight timeframe.
