Alessio Lomuscio is a professional at Safe Intelligence with education from the University of Birmingham.
Referenced in coverage
“Believes abliterated models can elicit useful behavior for stress-testing systems despite potential capability reduction.”