DeepSeek, a China-based AI startup, is temporarily halting new user registrations due to what it describes as "large-scale malicious attacks. " While users currently using its AI assistant have noted that it refrains from discussing politically sensitive issues in China, including the Tiananmen Square massacre, DeepSeek's recent announcement of its R1 model—claimed to offer performance comparable to OpenAI's ChatGPT at a significantly lower cost—has generated considerable interest, even propelling it to the top of the Apple App Store. The perception of DeepSeek as a competitor to U. S. AI companies triggered a sell-off of major tech stocks as the startup faced a cyberattack. The company informed users via its status page about the temporary registration limitations, assuring that existing users can still log in. By early Monday afternoon, DeepSeek experienced a partial outage of its web chat service and other login issues. It hasn't disclosed specific details regarding the cyberattack or the timeline for restoring sign-ups. The R1 model, released last week, is touted as being 20 to 50 times cheaper than OpenAI's equivalent model, raising industry eyebrows regarding its potential impact on U. S. AI firms that have heavily invested in technology development. Prominent tech figures like Jack Clark and Marc Andreessen have remarked on DeepSeek's significance as it challenges the dominance of Western AI companies.

However, user reports indicate that the chatbot avoids discussions on the Chinese Communist Party (CCP) and politically sensitive subjects. For instance, when asked about the Tiananmen Square protests or President Xi Jinping’s leadership, the chatbot redirected users by stating these topics were outside its current scope. Inquiries about the Uyghurs and Taiwan also highlight the chatbot's limitations. It initially began responding to sensitive topics but ultimately retracted, maintaining a narrative aligned with CCP views, such as stating Taiwan is an "inalienable part of China. " U. S. officials, including House Select Committee Chair John Moolenaar, have expressed concerns over DeepSeek, asserting that its AI model, under CCP control, minimizes historical atrocities and poses potential national security risks. They advocate for stronger export controls on technologies vital to DeepSeek's operations.