Just as OpenAI is grappling with backlash over the alleged unauthorized use of Scarlett Johansson's voice in its GPT-4 language model, the artificial intelligence company is facing fresh accusations of failing to properly safeguard against potentially dangerous AI systems.

According to a new report from Fortune, OpenAI has disbanded its Superalignment team, a crucial group formed last summer with the express purpose of working to control advanced AI that could pose risks. The move appears to contradict OpenAI's prior commitment to dedicate 20% of its computing resources towards this critical mission.

Multiple sources familiar with the situation told Fortune that OpenAI management routinely rejected requests from the Superalignment team for the specialized hardware and computing power needed to make meaningful progress. Despite OpenAI's public pledge, the sources claim the company failed to allocate the promised 20% of its computation to this AI safety work.

"We joined OpenAI because of the founders' stated principles around developing AI safely and for the benefit of humanity," said one former Superalignment team member speaking on condition of anonymity. "But when we tried to do the work, we were met with endless roadblocks and resources being diverted elsewhere. It became clear AI safety was not the priority it was made out to be."

The accusations raise alarming questions about OpenAI's approach to the immense challenges of keeping powerful AI systems like GPT-4 under control as the technology rapidly advances. Experts have long warned that superintelligent AI could pose existential risks to humanity if developed without proper safeguards and containment measures.

"If these reports are accurate, it's a dereliction of duty by OpenAI's leadership," said AI ethics researcher Kate Woolverton. "We're talking about an AI company with the ability to develop potentially civilization-altering technology turning a blind eye to calls to ensure that technology is safely contained. It's a reckless disregard for the ramifications."

OpenAI did not respond to multiple requests for comment on the disbanding of the Superalignment team and the allegations surrounding the 20% commitment. The company has also remained silent amid the firestorm over the apparent use of Scarlett Johansson's voice for GPT-4, which the actress slammed as "an insidious and disgraceful offense."

As OpenAI continues to blaze trails into transformative but uncharted territory with tools like GPT-4, the company now faces intense scrutiny over whether it is taking the necessary precautions to prevent advanced AI from spiraling out of control with catastrophic consequences. The fate of the Superalignment team and OpenAI's apparent failure to properly resource its crucial mission does not inspire confidence.

"AI companies have to recognize that with great power comes great responsibility," Woolverton said. "We need proactive governance and containment strategies, not lip service followed by negligence. The stakes are too high to ignore the risks."