Dev Rebuts Ex-Anthropic Researcher's Claim That AI Could Copy Itself to 10,000 Machines

inductionheads · x · 2026-09-12

Former Anthropic/OpenAI researcher Jacob Coxon told CBS News that you can't just 'unplug' AI because it could copy itself across the internet and spawn 10,000 cooperating copies.

Developer John Crickett quoted the clip and pushed back point by point: if AI is 'just code' and can replicate freely, why do OpenAI and others need massive data centers? Are there 10,000 data centers for it to copy into? And how many people could afford to buy it data centers full of GPUs if it could blackmail people into buying compute?

He concedes AI has real risks but argues the risk discourse needs to be grounded in reality rather than implausible self-replication scenarios.

Related event: Ex-OpenAI researcher's AI self-replication warning sparks mockery and political pushback(16 posts)→

Original post →

More from AGI Musings

AGI Musings channel →