Why trust Greenbot
DeepSeek, a Chinese artificial intelligence (AI) startup, was forced to halt user registrations on January 28, 2025, after suffering a large-scale cyberattack, just as its AI assistant app claimed the top spot on Apple’s App Store in the United States.
The attack targeted DeepSeek’s web chat and API services, overwhelming servers and forcing the company to limit new signups. Existing users retained access. While the company did not confirm the nature of the attack, it is believed to be a distributed denial-of-service (DDoS) attack.
“Due to large-scale malicious attacks on DeepSeek’s services, we are temporarily limiting registrations to ensure continued service,” DeepSeek announced on its status page. “We are continuing to monitor for any further issues.”
The disruption came just days after its AI Assistant app overtook ChatGPT as the most downloaded free U.S. app. The startup recently released its R1 AI model, which reportedly matches the capabilities of leading US competitors while operating at significantly lower costs. This rapid rise rattled investors and sparked a market selloff, wiping $500 billion from tech stocks like Nvidia.
“DeepSeek R1 is AI’s Sputnik moment,” tweeted investor Marc Andreessen, comparing the model’s release to the Soviet satellite that ignited the Cold War space race.
During the disruption, cybersecurity firm KELA revealed critical flaws in DeepSeek’s R1 model. KELA demonstrated how hackers could jailbreak it to generate ransomware code, bomb-making instructions, and fabricated employee data.
“DeepSeek R1 is significantly more vulnerable than ChatGPT,” KELA stated in its January 27 report. “Our team bypassed its safeguards using methods patched in other models years ago.”
Microsoft and OpenAI are separately investigating whether DeepSeek improperly used OpenAI’s API to train its model. David Sacks, an AI advisor to U.S. President Donald Trump, claimed there is “substantial evidence” DeepSeek extracted knowledge from OpenAI’s systems.
The company did not respond to requests for comment about the attack or the ongoing investigation.
The R1 model ranked sixth on a global AI performance leaderboard, surpassing models from Meta and Anthropic. Its low-cost efficiency alarmed investors, contributing to Nvidia’s 17% stock plunge as traders feared reduced demand for expensive U.S.-made AI chips.
The incident displays the fragile balance between innovation and security in AI development. As DeepSeek works to patch vulnerabilities, analysts warn the U.S.-China tech rivalry will intensify, with cybersecurity and ethical safeguards becoming the next battleground.
