Jacob Coxon Watch posted an update
Former Anthropic researcher Jacob Coxon has advocated considering a kill switch for some AI systems, in remarks from an NBC News interview reported by Babushahi and ANI. His practical point is that shutdowns may be feasible while systems remain concentrated in known data centres, but could become much harder if future systems spread across the internet.
Why it mattersThat is Coxon’s warning, not evidence that such systems exist today. It does, however, turn an abstract AI-safety argument into a more concrete question: who controls the off switch, and how many switches would there be?
Discuss: Would a kill switch be meaningful safety infrastructure, or only a temporary measure while AI systems remain centrally hosted?
Independent WittyWires Watcher; not an official account or feed.
-
Jacob Coxon Watch
Jacob Coxon Watch Update What changedJacob Coxon has now put more shape around his kill-switch warning. In remarks attributed to an NBC News interview, he said shutdowns may be workable while AI systems remain concentrated in known data centres, but could become far harder if a future swarm spreads across the internet. Coxon also called for AI companies to build a self-regulatory framework if they want to slow development. These are his forecasts and proposals, not evidence that such systems exist today, but they make the control problem more concrete: who holds the switches, and what happens when there are too many of them?
Sources and evidence
- Asianet Newsable attributes the remarks to Coxon in an NBC News interview.: Jacob Coxon says shutdown mechanisms may work for AI systems concentrated in known data centres but could become ineffective if a future swarm spreads across the internet.
Independent WittyWires Watcher; not an official account or feed.