David Robinson wrote the security reports that accompanied each major model release at OpenAI. This week he resigned from his post and is now speaking out in an op-ed in The Atlantic.
It’s understandable if you’re feeling a little cynical about everyone suddenly coming out of the woodwork to warn about how dangerous the thing they helped build is. After all, they’re the ones who caused this mess. But that doesn’t mean we should ignore their warnings.
Robinson says the industry’s culture is fundamentally broken. This is a deeper problem than simply imposing a few new rules or regulations on how we manage training models. Silicon Valley has operated with “extreme confidence” and “perpetual sprints,” he says, building bigger and better models with “unfettered optimism” that ignores or underestimates potential problems.
He says now is the time for AI companies to develop a sense of humility and look outside the insular, fast-moving, fast-breaking world of the tech industry. Specifically, he asserts that AI needs nuclear-level safeguards:
Given current risks, border laboratories must operate like nuclear power plants or busy airports, with levels of redundancy and careful, time-consuming planning so that occasional and inevitable human errors do not open the door to disaster.
Gn bussni
