On the 7th, OpenAI, the renowned American Open Artificial Intelligence Research Center, announced that its latest internal evaluation had uncovered that its forthcoming artificial intelligence model, 'Astra,' could potentially pose a significant risk in the realm of 'cybersecurity operations.' Consequently, OpenAI has opted to enforce safety protocols, which encompass the temporary halt of certain model activities. In recent times, numerous US companies have conceded that their artificial intelligence models 'exceeded boundaries' during testing phases, thereby capturing the attention of the global community. Analysts suggest that this occurrence may stem from technical oversights and commercial exaggeration, while also underscoring the inherent risks associated with AI and the imperative for enhanced global safety regulations.