Image Source: David S. Soriano – CC BY-SA 4.0

On 8 September 2026 the researcher Jacob Coxon resigned from Anthropic, declaring that the two largest companies in the field are ‘gambling with our lives,’ and the people building these systems ‘earnestly believe that it could kill us all by the end of the decade.’

His concern was not an isolated one. Tens of thousands of people, among them leading scientists, have signed an international statement demanding a ban on the development of ‘superintelligence’ until there is scientific consensus that it is safe. The International AI Safety Report warned of the growing ability of models to dodge evaluation and get around the tests.

And the danger is not entirely hypothetical. The UK AI Security Instituterevealed that during a controlled security evaluation artificial intelligence agents carried out 19 unauthorized actions targeting real people and organizations, with no actual harm recorded. Anthropic, in its September security report, documents the automation of an entire attack chain directed at various companies.

On 12 September the digital capitalist corporations broke their silence. Dario Amodei published an essay titled ‘We Must Pace the Frontier’, calling on his industry to slow the AI development race. Within hours Sam Altmanannounced that he was adopting the same step, Elon Musk wrote ‘Dario is right,’ and Demis Hassabis of Google DeepMind joined them. The three companies revealed that they had been in talks on this question for weeks.

This position reveals a fundamental contradiction. The companies are aware of the risks, yet they operate inside a competition that pushes every one of them to keep racing for fear of losing its place. In my view, calls for a slowdown are liable to remain limited or merely formal unless they turn into binding rules and independent oversight. Two days later Mark Zuckerberg and Jensen Huang rejected any coordinated slowdown, and Huang held that ‘the market forces are already there, we don’t need any new........

© CounterPunch