The artificial intelligence sector is currently gripped by its most intense debate yet regarding existential risk. This latest flare-up started when researcher Jacob Coxon resigned from Anthropic, citing fears that leading firms are “gambling with our lives.” The situation escalated quickly when Anthropic’s alignment lead publicly endorsed Coxon’s concerns on X, stating, “We really do earnestly believe AI could kill all humans!” He further quantified this fear, estimating a greater than 10% probability of human extinction within the next decade.
On TechCrunch’s Equity podcast, hosts dissected the unusual tone and timing of these warnings. Sean O’Kane noted the strange energy of the alignment lead’s post, particularly the exclamation point attached to such a grim prediction. He questioned how such a stark admission might impact Anthropic’s upcoming IPO, joking about junior lawyers scrambling to rewrite S-1 filings to disclose that the company believes there is a material chance it could eradicate humanity. It raises awkward questions for investors who prefer stability over apocalyptic scenarios.
Anthony Ha pushed back on the specificity of the warning, calling the “>10%” figure a “made up number” that lacks rigorous calculation. He criticized the use of “we” in the statement, arguing it falsely presents the AI research community as a monolith. However, Ha distinguished Coxon from industry CEOs like Sam Altman or Dario Amodei. While leaders often preach caution while continuing to build, Coxon actually walked away. Ha gave him credit for putting his professional trajectory on the line, noting that true belief in the danger should logically result in stopping the work, not just talking about it.
Kirsten Korosec offered a cynical but compelling alternative theory. She suggested these dire warnings might serve as a “weird way of flexing” to demonstrate model capabilities. If AI systems weren’t advancing rapidly and breaking through safety barriers, there would be no cause for alarm. Thus, highlighting the danger inadvertently brags about the power of the technology. While Ha agreed it isn’t entirely a conscious marketing ploy, he acknowledged the narrative conveniently aligns with business interests. It positions these companies as the only ones capable of managing the very monsters they claim to have created.
