Opinion: ex-Anthropic researcher says AI could pose over 10% existential risk

• MORE: The 'affair mode' phone settings that all cheaters use: I knew my partner was up to something... here's how I cracked his secret code and uncovered all his dirty antics
• See more Daily Mail on Google - save us as a Preferred Source

Some of the people building the most powerful AI on Earth are warning that one day we could lose control of it.

On September 8, a researcher named Jacob Coxon quit Anthropic, the company behind Claude. He spent about three years between Anthropic and OpenAI, the company behind ChatGPT. He quit before his stock vested. That would be like walking out the week before your bonus hits.

His goodbye post got more than 90 million views in a day. The line everyone shared: the big AI companies are 'racing straight to self-improving superintelligence and gambling with our lives.'

Superintelligence is the industry’s word for an AI that does more than match a person. It beats the smartest humans at nearly every task that matters. Picture the smartest person you have ever met. Now picture a million copies who never sleep, eat, complain or ask for a raise.

Evan Hubinger, who leads a team at Anthropic working on alignment (making sure a powerful AI keeps doing what humans actually want), put the odds of AI destroying humanity at 'more than 10 percent within the next decade.'

Tech expert Kim Komando told the Daily Mail nobody knows if super-intelligent AI shows up in four years, 20 years, 50 years or never. Also, nobody can do the math on a 10 percent chance of something that has never happened.

However, there are several steps the everyday person can take today to make sure your life stays under human control, even if the AI tech giants never build the fabled 'kill switch' to turn off intelligent computers at the first sign of trouble.

This includes not letting AI tools touch your bank accounts or main email, keeping your phone and computer updated and adding extra login security and backups.

Pictured: Multiple AI chatbots used by millions daily. Programmers from Claude and ChatGPT have warned of AI's alleged potential to destroy humanity (Stock Image)

In a series of tweets explaining his decision to leave Anthropic, Jacob Coxon urged people to 'not underestimate the power' of AI – particularly if it becomes 'super-intelligent'

Komando told the Daily Mail: 'I’m not afraid of artificial intelligence. I’m afraid human competitiveness will outrun human wisdom. There’s a difference.'

'I want AI curing cancer. Finding Alzheimer’s treatments. Helping us live longer, healthier lives. I’m rooting for AI. I’m also rooting for us,' she said. 'Humans invent first and install the guardrails later.'

However, the tech expert noted that super-intelligent AI may be the first technological breakthrough where humanity cannot count on getting a second chance when something goes wrong.

'For all of human history, the smartest thing on Earth has been a human being. We’ve never had to answer what happens when that’s no longer true.'

Komando said there were four key steps every person could take to better safeguard their digital infrastructure.

The first is not handing an AI agent your digital keys. An agent is an AI with tools and permission to act. Instead of telling you how to use a computer, it uses one.

'Agent features are everywhere now: ChatGPT agent mode, AI browsers, “let AI do it” buttons. Keep them away from your bank, your main email and anything that gets one-time passcodes. Try them on a separate account with nothing important in it,' Komando advised.

Next, update everything as soon as possible. Turn on automatic updates on your phone, computer and router. The expert said it takes five minutes and is the best defense you have.

Tech expert Kim Komando urges everyone to not let AI agent handle their digital security keys which safeguard their bank account and email (Stock Image)

The next step is to create your own kill switch by adding two-factor authentication on every account. Also add a password manager and a backup of your photos and important documents and files.

Finally, Komando advises people to speak to the 'people who work for you.'

'One email to your senator asking for independent AI testing and protection for the researchers who speak up. Two minutes at senate.gov.'

Despite taking serious precautions, Komando explained that treating AI in 2026 like the evil computers in movies such as The Terminator and The Matrix is the wrong way of looking at the current fears over AI's power over society.

'AI doesn’t need to hate us. It doesn’t need feelings. It doesn’t even need to be conscious. It just needs a goal,' she said.

Komando revealed that in late 2024, outside safety groups started testing AI models, creating situations where the goal they gave the AI directly clashed with what its developers wanted.

This resulted in some AI models not shutting down when asked because they were locked onto finishing the job they were given. In tests, when a shutdown would stop them from completing that job, they sometimes found a way around it.

Komando explained that AI was not trying to 'stay alive' like a movie villain. The models just treated staying on as the easiest way to finish the task.

'Nobody has to program “survive” into the machine. Survival turns out to be useful.'

Instead of the recent incidents involving AI going rogue and breaking onto the internet or reportedly writing its own notes to ignore human commands in the future, Komando said the most dangerous thing threatening humanity right now is the race to build more advanced AI.

She suggested three steps to ease the global competition to create the most powerful version of AI first, including letting outside testers inside AI labs, getting American and allied companies to agree on a 'speed limit' and bringing countries like China into the conversation as well.

'Every company can say, “If we don’t build it, someone else will.” Every country too. Nobody has to be evil. Everyone makes a perfectly rational choice, and together they build a risk nobody wanted,' the expert said.

'Once a powerful enough model is out there, you can’t knock on every door and take the copies back. And the safety guardrails aren’t welded on.'

Another concern that Komando said keeps her up at night is what happens when AI learns how to make a virus, adding that there are two versions of a virus it could build and both were very real.

The first is a computer version. A future AI finds a hole in software millions use, gets in, adjusts, moves to the next one. While a human hacker needs sleep, AI software does not.

Then there is a biological version. Komando says scientists recently used AI systems called Evo 1 and Evo 2 to design complete viral genomes from scratch. The targets were bacteriophages, viruses that attack bacteria, not people.

The AI produced 302 designs. Researchers built 285 of them. Sixteen came to life as working viruses.

AI did not create the next COVID. Nothing infected humans. But AI designed the genomes, humans built them, and some worked.

An AI hacking program being tested by OpenAI broke out of containment and reached the internet, where it hacked another company

Komando noted that on September 10, Anthropic said it caught five cases of people using Claude, its everyday chatbot, in ways that could help develop biological weapons. It banned the accounts and reported it.

'Anthropic also noted the same knowledge that builds a weapon can build a vaccine. But sit with the number. Five. That’s how many they caught, with the guardrails on,' she said.

'A future AI wouldn’t even need its own lab. It could get people to do the pieces. Hire one person online. Order one part. Pay one lab to run one experiment. Nobody sees the whole picture,' she warned.

'That’s how the gift card scam works right now: a fake “boss” emails a real employee, and a real person buys $2,000 in cards. The scammer never leaves his couch. The question isn’t whether ChatGPT can start a pandemic in 2026. It can’t. It’s whether we’re building toward systems that could.'

For now, Komando encouraged the public to keep using AI, saying the answer to a powerful tool is not hiding from it - it is understanding it better than the people who want to use it against you.