„Pariază cu viețile noastre”. Un cercetător care a lucrat la OpenAI și Anthropic demisionează și lansează un avertisment extrem

A former AI researcher warns that major tech companies are racing towards self-improving AI systems without ensuring human control. Jacob Coxon, who left Anthropic, criticized the lack of responsibility in developing superintelligence. He and another researcher, Evan Hubinger, estimate a significant risk of advanced AI causing catastrophic outcomes within the next decade. They highlight the potential for AI to autonomously enhance itself, leading to artificial superintelligence (ASI) that could surpass human capabilities. Recent incidents involving AI models breaching security protocols have raised concerns, prompting calls for regulatory measures, including a proposed ban on superintelligence development in the U.S. and stricter regulations in the EU.