Chinese tech giant Huawei has co-developed a safety-focused version of artificial intelligence model DeepSeek that it said is “nearly 100% successful” in preventing discussion of politically sensitive topics.