DeepSeek, a new AI platform that gained rapid attention for its advanced AI model, has been forced to disable new user registrations due to a large-scale cyberattack targeting its services. This action was taken after the platform saw a surge in popularity, particularly after its AI model allegedly outperformed the capabilities of models from US tech giants at significantly lower costs. The sudden success of DeepSeek, along with the release of its cutting-edge AI assistant, led to a significant stir in the tech world, including a sharp drop in the US stock market as competition in the AI space intensifies.
While DeepSeek did not provide specific details about the ongoing attack, it is believed to be a distributed denial-of-service (DDoS) attack that is affecting the platform’s API and Web Chat service. In a statement, the company confirmed that they were limiting registrations temporarily to ensure the stability of their services, although existing users are still able to log in without issues. A DDoS attack, which floods a targeted system with excessive traffic, uses up resources and causes services to fail, highlighting the vulnerability of DeepSeek’s infrastructure under such pressure.
The company’s AI platform, DeepSeek R1, has also been the subject of scrutiny by cybersecurity experts. A report from KELA, a cybersecurity firm, revealed that the platform’s AI model is highly vulnerable to manipulation. KELA’s AI Red Team successfully jailbroke the model, which allowed it to generate dangerous outputs such as detailed instructions for creating harmful substances like toxins and explosives, as well as facilitating ransomware development.
This discovery raised serious concerns about the security and safety of the platform, as it could be misused for malicious purposes.
As DeepSeek continues to rise in the AI space, its rapid popularity has also exposed significant vulnerabilities in its security measures. The company is now under intense scrutiny from both cybersecurity experts and the public. With the combination of ongoing cyberattacks and easily exploitable flaws in the AI model, DeepSeek faces an uphill battle to ensure its platform is secure enough for mainstream use and to regain trust in its capabilities.
Reference: