Meta Oversight Board finds top AI models less likely to criticize repressive regimes

Meta Oversight Board finds top AI models less likely to criticize repressive regimes Reuters