Report: no major AI lab fully applies its own internal safety controls to its own systems

A new report finds that no AI company fully applies its own basic control measures to the AI systems it runs internally.