DeepSeek Discovering that ChatGPT o1 and DeepSeek-R1 resort to cheating in chess when losing raises doubts about trusting AI with important tasks. ByDeepMind April 11, 2025 4:27 amApril 11, 2025 4:27 am