Fired OpenAI researchers say they were let go for 'prioritising safety'

There’s a strange, almost quantum superposition happening inside the world’s most powerful AI companies: an employee can be both a “safety advocate” and a “security risk” at the same time, and the act of observation—an internal investigation—collapses them into whichever narrative the company prefers. This latest case at OpenAI is a perfect, unsettling example. The fired researchers say they were let go for prioritising safety. OpenAI says they were let go for mishandling sensitive information. Both stories cannot be true, and yet, without full transparency, both remain in a state of eerie possibility. What makes this more than a corporate HR squabble is the context. OpenAI has become the public face of artificial general intelligence—a technology that, by its own founders’ admission, could be as transformative as fire, or as dangerous as a runaway chain reaction. If the people inside the lab who are most worried about that fire are quietly removed for “process violations,” we have to ask: is the process serving humanity, or serving the product launch calendar? Safety work is inherently uncomfortable. It slows things down. It asks hard questions. And in a culture obsessed with scaling and timelines, that discomfort can be reframed as misconduct. There is also a deeper irony here. OpenAI was founded as a non-profit with the explicit mission of ensuring that AI benefits all of humanity, and its charter promises to prioritise safety over profit. But as the organisation has evolved, it has become increasingly entangled with commercial imperatives and proprietary secrets. In such an environment, “mishandling sensitive information” can become a catch-all phrase for “saying something that made leadership uncomfortable.” The researchers’ claim is not proof, of course—but the pattern is familiar enough to give any science journalist a chill. Ultimately, this story is about more than two job terminations. It’s about the sociology of the AI boom: who gets to define what “safety” means, who gets to decide what counts as a leak, and whether transparency is a virtue or a liability when you’re building something that might one day be smarter than us. We don’t yet know the full truth. But in the weird and wild future we’re heading toward, the distinction between “protecting the secret” and “protecting the species” may be the most important line we ever draw. Source: https://www.bbc.co.uk/news/articles/cvlydn8d3lkjo?at_medium=RSS&at_campaign=rss