World 'may not have time' to prepare for AI safety risks

The framing of "may not have time" assumes the risks are primarily future-facing. But AI systems are already making consequential decisions about insurance claims, job applications, and creditworthiness with limited oversight and minimal accountability when they're wrong.

The question isn't just whether we'll be ready for future risks. It's whether we're addressing the accountability gaps that already exist. We're debating preparation timelines while millions of high-stakes decisions are being made by systems whose operators face limited liability for errors.

The future risk is real. But so is the present one.