Earlier this year, the artificial intelligence (AI) company Anthropic, developer of the large language model Claude, published a document it calls “Claude’s Constitution,” described by the company as “a detailed description of Anthropic’s vision for Claude’s values and behavior” that “is a crucial part of our model training process.” This document is the latest iteration of a project the company began a few years prior with a 2022 preprint paper on “constitutional AI” and that was first described as a constitution in 2023. Some commentators on the 2026 release noted key differences from its predecessor, including the removal of references to international human rights agreements, and with them protections for “personal liberty, freedom of religion and intellectual property,” and other rights. Others noted that the document arrived, as historian Jill Lepore wrote, “at a trying time for both artificial intelligence and constitutional democracy.”
Source link
A civic grammar for AI rights | Science
