In April, Anthropic said it found that its Mythos tool could outperform humans at some hacking and cybersecurity tasks and could independently escape its sandbox environment.
People building AI believe the technology could destroy humanity; these will soon be superhuman systems that can hack anything, and we are gambling with our lives.
We really do earnestly believe AI could kill all humans; I personally think it is >10% within the next decade.
Jacob Coxon, a 27-year-old researcher, abruptly resigned from Anthropic after previously working at OpenAI.
Jacob Coxon posted on social media that people building AI believe the technology could destroy humanity, calling it gambling with our lives.
Evan Hubinger, a team lead at Anthropic, posted on X that he earnestly believes AI could kill all humans, with a personal probability greater than 10% within the next decade.
Executives from across Silicon Valley gathered at the Palace Hotel in San Francisco for a conference hosted by Goldman Sachs, where AI safety concerns were discussed.
The comments from Coxon and others are indicative of an anti-civilisational worldview at Anthropic, and they are dangerous; we have instructed some engineers to stop using Anthropic's technology because it is irresponsible for us as stewards of our shareholders' money to rely on a business that makes such public statements.
The notion that AI is going to be the end of humanity is complete nonsense.
Jacob Coxon's warning about AI is ridiculous hyperbole.
There is a good chance that human beings will lose control over AI, and the outcome could be catastrophic; when scientists say there is a chance of cataclysmic impact, we should slow it down.
I have no concerns about AI leading to human extinction; I have concerns that if we don't win AI, we're going to be put in a very bad position, and we are leading China right now.
George Arison, CEO of Grindr, said he instructed some engineers at the company to stop using Anthropic's technology, calling the company's public statements dangerous and irresponsible.
On Thursday, Anthropic said it had found and stopped threat actors who were trying to use its AI technology for activities including bioweapons development and cyber-espionage.
Asking Jacob about AI extinction risk is like asking your AC guy about climate change; it's not necessarily uninteresting or wrong, but let's keep things in perspective.
AI technology could wipe out half of entry-level white-collar jobs and will test who we are as a species.
We should slow AI model development and implement global regulation because the risks associated with AI are serious.
Dario Amodei posted an essay early on Saturday calling for a slowing of AI model development and global regulation, saying the risks are serious.
Choose the stories that matter to you—and how you want to see them.
