Allowing independent hackers to legally circumvent digital security measures to probe artificial intelligence models for bias and discrimination would boost transparency and trust in the nascent, polarizing technology, policy groups told the US Copyright Office.