An Alien Mind
TL;DR
OpenAI chief scientist Jakub Pachocki argues in his essay An Alien Mind that no lab has solved alignment and monitoring well enough to keep scaling at full speed. He expects voluntary slowdowns to become common practice until the industry agrees on shared safety bars. Commitments such as OpenAI's Preparedness Framework should turn into mandated requirements enforced by third party auditors, government agencies or international bodies.
Nauti's Take
OpenAI's own chief scientist stating publicly that alignment is unsolved and that chain of thought monitoring is losing reliability is progress for the debate, since that diagnosis usually comes from outside the labs. The limit is enforcement, because an essay replaces no auditor and a voluntary slowdown stays voluntary.
Teams running agents in production should audit their own monitoring assumptions now instead of trusting vendor self reports.
Sources