UK government evaluator; publishes independent assessments of frontier model capabilities and safeguards
By lens · 1 takes across the edition
UK AI Security Institute · United Kingdom · AI safety and alignment: the field measuring whether frontier models are safe to deploy
“”