On Sept. 8, 2026, former Anthropic and OpenAI employee Jacob Coxon quit his job at the artificial intelligence (AI) companies, making a detailed chain of posts on explaining his reasoning. Coxon brought forth the concern that AI poses a genuine existential threat to humanity, a view he claims his former colleagues shared, posting, “The people building AI earnestly believe that it could kill us all by the end of the decade.”
While many have taken Coxon’s claims as serious declarations of imminent danger for humanity, he has also faced critique and suspicion from many who believe his statement to be a marketing stunt by Anthropic. According to , a similar stunt was pulled by the company earlier this year when their newest model at the time, Mythos, was purported by many key figures in the industry to be far too capable of breaching security and publicizing private information.
The reports were later accused by many, including cURL developer , of being primarily designed to produce hype around the model’s supposedly advanced capabilities, while remedying any security concerns with the public release of the reportedly downgraded model Fable. While this event and in which AI companies have been accused of using public statements of danger to generate hype around AI capabilities have understandably led to suspicion surrounding Coxon’s statement, one must ask if the statement holds any real weight, regardless of the intent behind it.
Certain assertions within Coxon’s post carry an apparent weight that was not present in the warnings presented about Mythos. While Mythos purported to pose security and informational concerns, Coxon makes graver predictions about the possible implications of AI’s advancement.
“Do not underestimate the power of this technology,” Coxon said. “These will soon be superhuman systems that can hack anything, revolutionize any field overnight, and acquire real power and resources.”
Statements such as these, alongside the previously mentioned mortal threat which Coxon claims AI poses, seem as though they may be a step too far for a marketing stunt. In addition to this, many experts in the field seem to share many of Coxon’s concerns. According to , Toby Ord, a senior researcher at Oxford’s AI Governance Initiative, claimed: “Just as we can’t uninvent the bomb, once we develop AI systems that surpass humanity it would be almost impossible to go back.”
On the other hand, by incrementally increasing the quantity and intensity of these claims, AI companies continually generate higher and higher expectations of the capabilities of their products to an increasingly desensitized audience. The desensitization that results from a gradual raising of the stakes could be seen as highly beneficial to companies such as Anthropic. Critics of artificial intelligence have raised a great number of legitimate involving AI and its continued development. Narratives like Anthropic's portray the dangers of AI as if we are on the cusp of experiencing the plot of a science fiction movie such as the 1983 film “WarGames,” which has the effect of sensationalizing the threats AI poses, while distracting from these concerns.
So does artificial intelligence really pose imminent mortal danger to the whole of humanity as Coxon purports? In my opinion, it very well could, but this is something that we have known for a great length of time before Coxon’s statement, before the publicity around Mythos and Fable, and before any other modern incident in which the possibility of AI gaining the ability to hack, expose private information or cause wars had been discussed. We are seeing the negative effects that AI and the construction of data centers are having on the environment and on marginalized, low-income and environmentally vulnerable populations right now. According to the World Resources Institute, a national review surveying about 700 data centers across the United States found that close to half of them had been built near or within communities dealing with dangerous environmental factors such as water and air pollution. In South Carolina, many new data centers—as well as the fossil fuel generators that are used to power them—are built near low-income and predominantly Black communities. This is problematic because of the strain that data centers place on the water supplies of the areas where they are built. states that medium-sized data centers can use around 300,000 gallons of water daily, while larger ones can use up to 5 million. The existence of data centers often places great environmental strain on areas that are already struggling in terms of health, income and acquisition of essential resources like water.
I do not believe I can answer the question of whether or not Coxon’s statement was a part of a marketing stunt, nor can I confidently make the claim that AI will soon have the capabilities to start a nuclear war, but I can assert that we do not need to look to the future in order to see the dangers that AI poses to human life and wellbeing right now. I believe that AI restriction and regulation are imperative, and not only under the threats that figures like Coxon bring forward as possibilities, but under the very real threats that AI poses to vulnerable populations right now.
The Student Movement is the official student newspaper of 星空无限传媒. Opinions expressed in the Student Movement are those of the authors and do not necessarily reflect the opinions of the editors, 星空无限传媒 or the Seventh-day Adventist church.