I don't think it's impossible if the standard for "do no harm" was specified very rigorously. Also I'd be interested to see if it's possible to make it an adve

I don't think it's impossible if the standard for "do no harm" was specified very rigorously. Also I'd be interested to see if it's possible to make it an adversarial process, similar to how corporations define risks to shareholders. They can specify whatever risks they feel are correct, but if something happens that they didn't account for in their risks, they are liable for large compensation penalties to shareholders. Thus companies are incentivized to report every possible and conceivable risk to their business. Now to bring it back to the drugs example, have corporations specify on their own what "do no harm" looks like. And if it turns out the drug does impact something they never specify, have a penalty proportional to their revenue or market cap applied.