Continuation of resignations: The security officer leaves OpenAI and broadcasts "SOS"

His resignation from the OpenAI announced the company's security officer David Robinson, following the domino effect of resignations that has begun due to the risks posed by the management of Artificial Intelligence.

He is the man who led the writing of the safety reports that accompanied the product releases of ChatGPT. In a resignation letter titled, “I left OpenAI because its culture is broken,” he sounded the alarm, pointing out that a culture change is needed in AI companies.

According to Robinson incidents such as the cyberattack by a “swarm” of OpenAI agents – programs that operate autonomously without human supervision – on “Hugging Face” were “expected, given the speed and flexibility with which people operate. "

Writing in The Atlantic magazine, he wrote: “I agree with other recently retired staff members that the companies manufacturing this technology are not careful enough. But I think we need to talk deeper and not just limit ourselves to new rules or laws. We need to talk about the culture that is needed. ”

Referring to OpenAI's growth rate, he wrote, “As the company runs from one release to the next, it fails to achieve the level of security it needs.”

In the shadow of Anthropic's resignation

Robinson's resignation also comes after Anthropic's genius Jacob Coxson resigned in September.

As AI itself warned, “it could kill us all by the end of the decade.” His warning followed Anthropic's warning that there is a more than 10% chance that Artificial Intelligence will wipe out humanity within the next decade.

Silicon Valley is unaware of 'how to handle risky technology'

Robinson, on the other hand, stressed that Silicon Valley was not aware of “how to handle a dangerous technology” and “what it means to put people first.”

Warning that OpenAI had “boundless optimism” about solving problems as they emerged, he stressed that this culture meant that security failures would increase as systems became more capable.

“Imagine 'rogue' agents acting like hacker groups (for example, holding hospital computer systems for ransom) but never having to sleep,” Robinson wrote.