In a recent LinkedIn post, Linas Beliūnas highlights significant concerns raised by AI researchers regarding the rapid advancement of artificial intelligence and its potential existential risks. Beliūnas shared news about a former researcher from OpenAI and Anthropic who publicly resigned, warning that these leading AI companies are accelerating towards superintelligence and, in doing so, are “gambling with our lives.”
Beliūnas brought attention to the specific claims made by AI researcher Jacob Coxon. According to the insights shared by Beliūnas, Coxon stated that companies such as OpenAI and Anthropic are aware of the profound risks associated with advanced AI, including the possibility of systems that could lead to humanity’s destruction by the end of this decade. Coxon’s perspective, as relayed by Beliūnas, suggests that these organizations feel compelled to continue their rapid development because the competitive landscape means no single entity can afford to slow down.
“AI researcher Jacob Coxon said companies like OpenAI and Anthropic know the risks, including systems that could destroy humanity by decade’s end, but feel trapped in a race where others won’t slow down.”
The Urgency of Internal Warnings
Further amplifying these concerns, Linas Beliūnas pointed to the agreement from Evan Hubinger, Anthropic’s current Head of Alignment Stress-Testing. Hubinger reportedly confirmed the gravity of the situation, with Beliūnas relaying his personal assessment that there is a greater than 10% chance of AI causing humanity’s extinction within the next ten years. Crucially, Hubinger also confirmed that Anthropic has not yet developed a definitive plan to mitigate this existential threat.
Beliūnas emphasized the importance of heeding such warnings from individuals deeply involved in AI development. As Linas Beliūnas noted in his post:
“When the people closest to the models start ringing alarm bells, it’s time to listen.”
Broader Implications for AI Development
The insights shared by Linas Beliūnas underscore a growing unease within the AI research community itself. The fact that researchers from prominent organizations like OpenAI and Anthropic are voicing such stark warnings suggests that the pace of AI development may be outpacing safety considerations.
The Competitive Race Dynamics
Beliūnas’s post highlights the ‘race dynamic’ as a significant factor. The competitive pressure to achieve breakthroughs in AI, particularly in the pursuit of superintelligence, appears to be creating an environment where safety protocols might be secondary to speed. As Beliūnas relayed from Coxon’s statement, this creates a situation where companies feel they ‘can’t slow down’ even if they recognize the potential catastrophic outcomes.
The Lack of a Solution
Another critical point Beliūnas brought to light is the admission regarding the absence of a clear solution for AI alignment and safety. Evan Hubinger’s confirmation that Anthropic ‘still does not have a plan to solve the problem’ is a particularly sobering revelation, as reported by Beliūnas. This suggests that the very organizations at the forefront of AI innovation are grappling with fundamental challenges in controlling advanced AI systems.
In conclusion, Linas Beliūnas’s LinkedIn post serves as a crucial disseminator of internal alarms from within the AI industry’s leading institutions. The concerns raised by researchers about the race towards superintelligence and the lack of safety plans warrant serious attention from policymakers, industry leaders, and the public alike.
P.S. Beliūnas also shared a link to an article about Grok Bot, noting it as a more positive development in the AI space.
📝 About This Content
This article is based on insights shared by Linas Beliūnas on LinkedIn.
📅 Originally posted on September 9, 2026 | View original post on LinkedIn →