The Commission on Human Rights (CHR) is calling for a broader right to human review of automated decisions, saying the proposed artificial intelligence (AI) law could leave people without an effective way to challenge errors or discrimination.
In a position paper, the CHR said the House substitute bill for the proposed Artificial Intelligence Development and Regulation Act would limit the right to human review to automated decisions that have a legal or similarly significant effect on a person.
Section 6.5 of the substitute bill sets out the right to human review and addresses when that right may be restricted.
The CHR recommended extending that right to “any automated decision that affects the rights, interests, or legitimate expectations of a person.”
“Artificial intelligence and automated systems are increasingly used in decisions affecting employment, education, financial services, healthcare, social protection, access to public services, and other opportunities,” the commission said.
According to the CHR, automated decisions can have substantial consequences even when they have no immediate or formal legal effect.
Under the proposal, a human reviewer should be able to assess the outcome independently and correct or reverse it when warranted.
The position paper also calls for an “adequate and comprehensible explanation of the automated decision, including its principal basis and the relevant factors considered in reaching the outcome.”
Such an explanation, the commission said, should give people enough information to challenge the decision without requiring disclosure of proprietary source code or trade secrets.
The CHR further proposed allowing limits on human review only when strictly necessary and proportionate to achieve a legitimate objective.
Moreover, the commission backed passage of a comprehensive AI law and also called for explicit protection against privacy infringements by government and private actors.
On AI-generated sexual content, the CHR urged lawmakers to remove an exemption for bona fide satire and parody from the provision prohibiting nonconsensual intimate synthetic media and AI-generated child sexual abuse material.
The exemption, the commission warned, could allow AI-generated intimate content to spread and cause harm before authorities determine whether the content qualifies as bona fide satire or parody.
Furthermore, the commission also urged lawmakers to define the reasonable-person test used in that provision to determine whether AI-generated content realistically depicts an identifiable person in a sexually intimate manner.
Under the CHR’s proposal, the test would consider the depiction’s appearance, the context in which it was created or shared, and any available identifying information.
Follow Tan Briones & Associates on LinkedIn for more legal updates and law-related articles.
