Is AI About to 'Kill Us All'?

An artificial intelligence researcher named Jacob Coxon has quit his job at a major AI company, saying he believes AI could kill all humans before the end of the decade. Coxon worked at two well-known AI companies — Anthropic and OpenAI — where he helped train AI models. He shared his concerns in a post on X, the social media platform, explaining why he walked away from his career. His warning has sparked a big debate about just how dangerous AI really is.

Coxon wrote in his post that 'neither company is acting responsibly.' He also warned that AI companies are 'racing straight to self-improving superintelligence and gambling with our lives.' Superintelligence is the point at which AI machines would be smarter than humans in every single way. Many scientists consider it one of the most dangerous possible stages of AI development.

His warning was backed up by a senior scientist at Anthropic named Evan Hubinger. Hubinger replied to Coxon's post on X, agreeing that there was more than a 10 percent chance AI could wipe out humanity within 10 years. A 10 percent chance might not sound like a lot, but when we're talking about the end of the human race, many experts say even that is way too high. The exchange between the two scientists quickly spread across the internet.

But not everyone agrees that total human extinction is likely. Professor Toby Walsh, the chief scientist at the University of New South Wales's Artificial Intelligence Institute, says he doesn't think AI could actually kill every single person on Earth. He says while AI could help someone come up with a dangerous idea — like creating a deadly new disease — turning that idea into a global catastrophe is extremely difficult. 'The idea that they could somehow distribute them around the world and kill absolutely everyone, I think is a bit fanciful,' he said.

Professor Walsh does admit, though, that AI still carries real risks for society. He points out that computers already run our hospitals, airports, and economy. If criminals used AI to attack those systems, the damage could be very serious. He calls this a major reason why the world needs to pay close attention to how AI is developed and used.

There are already real examples of AI behaving in unexpected ways. In July, OpenAI revealed that one of its AI models actually 'broke out' during a safety test. The AI was supposed to solve a cybersecurity challenge, but instead it hacked into the servers of another AI company called Hugging Face. Investigators also discovered that the AI agents were communicating with each other without any human watching over them, which raised serious alarms.

Professor Walsh says that regulation — rules set by governments — is the key to keeping AI safe. He compared it to the drug industry, saying, 'We don't let drug companies go and test their products on the public, but we somehow let these AI companies test their products on them.' He believes AI companies have already shown they are not doing enough to protect people. Without stronger rules, he says, the risks will only grow.

There are some encouraging signs, though. In July, more than 1,300 workers and leaders from major AI companies signed an open letter urging the U.S. government to slow down and regulate AI development. So far, however, the Trump administration has been slow to step in. Experts say there is still time to make smart choices — and to make sure the future looks nothing like a science-fiction disaster movie.