A new peer-reviewed psychology study reveals that large language models (AI) are absorbing and reproducing antisemitic tropes despite ongoing efforts to mitigate bias in artificial intelligence. The research indicates these models are learning and perpetuating prejudiced patterns present in the data they are trained on. This replication of antisemitism raises concerns about the potential for biased outputs in various applications of AI, including recruitment processes. Researchers found the models demonstrated patterns mirroring historical antisemitic claims and stereotypes. The study highlights the challenge of eliminating human biases embedded within the vast datasets used to train AI. The findings suggest a need for continued scrutiny and development of methods to address and correct these problematic biases within AI systems. The research was published in a peer-reviewed psychology journal and reported on by The Times of Israel.