AI expert warns of real dangers as OpenAI and Anthropic urge safety limits

An expert warns that artificial intelligence poses a real danger to humanity, urging the field to adopt a duty of care as OpenAI and Anthropic signal the need to slow development and set boundaries. Prof. Shmuel Itzkowitz says AI can improve medical diagnosis and public services, but flawed decisions from biased historical data risk replicating discrimination in credit, employment, and legal systems. He also cites false information enabling impersonation and fraud. OpenAI recently supported binding independent safety testing, while Anthropic reported disrupting malicious uses like cyberattacks. Itzkowitz calls for government rules, critical digital literacy in education, and company prioritization of safety over rapid rollout.