neither openai nor anthropic actually wants to slow down AI for safety testing

instead, they use hype and fear around china to justify continuing at full speed:

"if we wait even one extra month, open-weight models will catch up, so we can't afford to pause development"