DeepSeek outage adds to growing pains amid political scrutiny, but some see opportunity

Days after allegations of a cyberattack against the hot Chinese artificial intelligence (AI) start-up DeepSeek, the company continues to face service issues, while the broader industry and national governments debate whether to shun or embrace the company’s cost-effective technology.

Some users on X, formerly Twitter, continued to report issues in accessing DeepSeek’s self-hosted AI service on Friday, hours after the company said it had rolled out a fix for a massive distributed denial of service (DDoS) attack. The company issued a notice at 1.19am China time on Friday saying a fix was implemented and that it was monitoring the results. Yet it continued to limit new sign-ups to Chinese phone numbers, email addresses or Google sign-ins, and warned visitors to its website that “registration may be busy” because of “large-scale malicious attacks on DeepSeek’s services”.

A report from Chinese state broadcaster CCTV this week alleged that the attacks originated in the US. Some unverified claims online said the DDoS, which overwhelms a server with requests in rapid succession, reached as high as 230 million requests per second.

Hangzhou-based DeepSeek is facing growing pains from its rapid rise this month as its latest open-source models, V3 and R1, received global attention for matching the performance in some benchmarks of the closed-source GPT models from industry leader OpenAI. Accompanying its new-found fame is unprecedented legal scrutiny, with the US alleging that DeepSeek stole OpenAI data and other countries seeking to curb access to DeepSeek servers because it routes data to China.