Teaching AI teams to report unsafe tasks
A new study shows how to design audits and training so AI agents reliably flag unsafe tasks instead of staying silent.
A new study shows how to design audits and training so AI agents reliably flag unsafe tasks instead of staying silent.