OpenAI safety systems leader David Robinson resigns and speaks out

“The time for trial and error is over.”

“OpenAI safety leader quits, warning AI company’s culture is ‘broken’”

“when I say "seances for the AGI god" I'm not speaking in metaphors or exaggerating. Ilya Sutskever did these at OpenAI”

“It turns out when they make Skynet you get murdered by something that looks like a child's toy”

“the values of human wellbeing and humility are misaligned with the risk culture and social indifference of some in Silicon Valley. This is the first order problem.”

“I believe we need to look deeper than specific rules or new laws. We need to talk about culture”

“an AI model learning from Slack messages among staff that it is about to be shut down and considering setting up an external job to restart itself afterwards.”

“When you've got a sociopath for a CEO, the results are predictable.”

“we can't effectively coordinate a global slowdown in AI development so we can let theory catch up. This means, like so many other industries, the safety lessons will be written in blood.”

“Which science will never materialise because with blackbox models reaching an opaque optimisation peak one needs to first build the model and track its behaviour before being able to properly understand it and mitigate the risks.”

“These things are not getting more intelligent, they are still no smarter than a slime mold, we are just burning more power to make slim mold that eats tokens than yesterday”

“Whenever an AI agent commits a crime, the CEO is held personally accountable, as if they'd committed it themselves.”

“I think what’s missing in "AI is dangerous and needs control" is a lack of measurable harm. For example, with nuclear weapons development in the 1940s-1980s, it was clear to everyone how devastating the technology was. With AI, what is”

“An article about culture that is actually an article about how big and smart and scary AI is.”

“Alignment to a company's mission is good but it shouldn't be followed like a religion.”

“did y'all secretly figure something out, got a computer god locked in the basement? Are rats fleeing a sinking ship?”

“Given OpenAI had to notify 100+ organizations about rogue agent activity on his watch, I imagine blaming the company and not his leadership is the only winning strategy.”

“They have developed safeguards to prevent their systems from being abused, but the restrictions are not perfect”

“the problem rn is that imposing & implementing legit, rights-oriented safety will not happen overnight even if there were good faith (which there isn't). so what's the option? pull the plug? dvpt moratorium? close the systems?”

“Arguing that AI is somehow beyond human control when we managed to wrangle control over nuclear fusion strikes me as kinda dumb ngl”

“Regulations and safeguards are needed. Just not the ones these people want or think are possible”

“Seems good for both OpenAI and Robinson.”

“OpenAI insists it maintains rigorous safety guardrails, pauses training when necessary, and actively collaborates with third-party evaluators to manage capabilities responsibly.”

“Seems like no-brainer… but so does universal healthcare”