AI develops bias differently than humans
Researchers at MIT have published findings showing that AI hiring tools do not simply learn stereotypes from training data — they develop biased patterns through mechanisms distinct from human prejudice. The study tested multiple commercial AI recruitment platforms and found that they produced discriminatory outcomes against candidates based on gender, race, and age at rates exceeding human hiring managers in controlled comparisons.
The research team warned that because AI systems process thousands of candidate features simultaneously, identifying and removing biased correlations is far more complex than addressing human prejudice in interviews.
Anthropic settles copyright lawsuit for $1.5 billion
In other AI news, a U.S. judge approved Anthropic's $1.5 billion settlement of a copyright lawsuit brought by authors who alleged the company's AI models were trained on their works without permission. The settlement is among the largest in the AI industry and sets a precedent for how AI companies compensate content creators whose material is used in training datasets.
The case has been closely watched by the technology sector as legal challenges over training data continue to mount against major AI developers.