Former Anthropic researcher Jacob Coxon has publicly criticized the company's leadership for its aggressive approach to artificial intelligence development. During a recent Q&A session on X, Coxon argued that the company's reliance on consequentialist philosophy to justify pushing the boundaries of AI research is fundamentally flawed.
Coxon specifically highlighted the company's focus on recursive self-improvement, a process where AI models build their own successors. He claims this strategy forced competitors like OpenAI to abandon projects such as the Sora model to keep pace with Anthropic's rapid development cycle.
The former researcher also noted that Anthropic leadership maintains deep skepticism regarding potential negotiations with the US and Chinese governments. Coxon believes that if an AI arms race is viewed as inevitable, companies should choose not to participate in it at all.
In response to such concerns, Anthropic CEO Dario Amodei has stated that slowing down development carries its own risks. He told CNN that if his company decelerates, there is a significant danger that the technology could eventually be controlled by entities with inadequate safety standards.
