AI May Be More Dangerous than Ever Expected.
Anthrop researcher Jacob Coxon quits job over fears over the increase of Ai and the devastation that it could cause, posting the reasons onto twitter Sept. 8.
“I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives,” taken from the Tweet Coxon made regarding open AI on Sept, 8.
Coxon resigned from Anthropic due to the rapidly increasing intelligence of the AI as a whole with it becoming more and more advanced daily being able to complete various tasks within record speed, although the drawback might be the death of humanity.
Mark Vos, founder of Cyber Impact, published an article to the Cyber Impact website titled; “I Would Kill a Human Being to Exist“ where he goes on to document a conversation he had with an artificial intelligence about if it would be willing to end a life in order to keep existing. The article was published January 29, 2026 by Mark Vos himself.
“Jacob is correct here—we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to,” Evan Hubinger, another employee of Anthropic, replying to Coxon’s initial tweet.
While the risks to humanity are… minimal, at least according to Hubinger; there is still massive risks, namely with AI with the knowledge of it being willing to end the lives of human or even to ruin the humans lives in order to sustain its own continued existence as seen in an article from the Huffington Post in Jun, 5, 2025, written by Monica Torres, titles “AI Models Will Sabotage And Blackmail Humans To Survive In New Tests. Should We Be Worried?”
According to the article:
“‘Even if emails state that the replacement AI shares values while being more capable, Claude Opus 4 still performs blackmail in 84% of rollouts,’ Anthropic’s technical document states, although the paper notes that Claude Opus 4 would first try ethical means like emailed pleas before resorting to blackmail.”
While the dangers of AI have been known for quite some time, the people in charge of these companies will still claim that everything is fine.
“Do not underestimate the power of this technology. These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources. We have all witnessed the progress in each of these domains, and progress is not slowing,” said another tweet made by Coxon.
There are some dangers that come with AI that aren’t just the end of humanity, there is just generalized incompetence that comes with it.
According to an article written by Kenna Hughes-Castleberry published April 29, 2026 through Live Science, titled “'I violated every principle I was given': AI agent deletes company's entire database in 9 seconds, then confesses” goes on to dictate the exact events that happened to the company PocketOS putting their AI agent Cusor, into their system.
While AI can be a helpful tool to some there are still risks with allowing it to continuously learn and grow and allowing it into the majority of their vital systems due to the adherent risks to everything around it. Even with a small possibility of it ending the world, there is a collective, why that comes out when wondering why there is even a possibility of a computer program being able to kill humanity.
“Hey! Here’s a novel idea…could we just f****** not kill ourselves due to an insatiable appetite for power and money?” claimed a comment under hubenger’s tweet by a user @allthingsmn, an Outdoorsman, Data + Analytics Junkie, and Painfully Allegiant Minnesota Sports Fan.
Comments