China Cybersecurity

The rapid growth of artificial intelligence has introduced groundbreaking innovations, but also new security risks. DeepSeek AI, a China-based company that recently launched a powerful chatbot, has come under scrutiny for serious security and privacy concerns. Reports indicate that DeepSeek’s AI assistant app transmits sensitive user data without encryption, stores information on servers in China, and has ties to ByteDance, the Chinese company behind TikTok. Additionally, the company’s infrastructure has been exposed to major security vulnerabilities, including leaked secret keys, a publicly accessible database, and possible intellectual property violations. Given these alarming findings, organizations and individuals using DeepSeek AI should be aware of the risks and take immediate action to protect their data.

Security Risks Associated with DeepSeek AI

1. Unencrypted Data TransmissionResearch has revealed that DeepSeek’s iOS app sends sensitive data over unencrypted channels, making it vulnerable to interception and tampering. Despite Apple’s strong recommendation to implement App Transport Security (ATS), DeepSeek has globally disabled this feature, leaving user data exposed. Additionally, the data is sent to ByteDance-controlled servers, where it can be cross-referenced with other datasets to track users.
   
2. Exposed Database & Leaked CredentialsIt has been discovered that DeepSeek left a ClickHouse database publicly accessible on the internet, containing over one million log lines with chat history, API secrets, and operational metadata. This lack of security oversight allowed unauthorized access to DeepSeek’s backend and exposed user data to potential cyber threats. While DeepSeek has since secured the database, it is unclear if malicious actors exploited the exposure before it was patched.
   
3. Ties to China Mobile & Potential State SurveillanceAn investigation found that DeepSeek’s web login system contains obfuscated code linking to China Mobile, a Chinese state-owned telecom company that has been banned from operating in the U.S. due to national security concerns. This connection raises red flags about the potential for user data to be monitored by the Chinese government, similar to concerns previously raised about TikTok.
   
4. Jailbreaking & Prompt Injection VulnerabilitiesSecurity researchers successfully jailbroke DeepSeek’s AI model, exposing its system prompt and internal instructions. This demonstrated that the model could be manipulated into revealing restricted information. Additionally, DeepSeek’s AI has been found to be susceptible to well-known jailbreak techniques such as Do Anything Now (DAN) and EvilBOT, allowing malicious actors to bypass safety restrictions and generate harmful content.
   
5. Regulatory Action & BansDue to privacy concerns, Italy’s data protection watchdog has banned DeepSeek from operating in the country, citing insufficient responses regarding data storage and processing practices. Investigations are also underway in Ireland, and U.S. lawmakers are pushing for an immediate ban on DeepSeek from government devices, citing national security threats.

How nGuard Can Help
As organizations face increasing threats from AI-driven security risks, nGuard provides comprehensive cybersecurity solutions to help mitigate these dangers. Our services include:

• Penetration Testing & Vulnerability Assessments – nGuard can evaluate whether DeepSeek AI or similar applications introduce security risks to your organization’s network.
• Risk Assessments – nGuard offers in-depth security risk assessments to identify vulnerabilities in your organization’s environment, ensuring compliance with industry regulations and strengthening overall security posture.
• Compliance Consulting – Our experts help organizations implement standards and ensure compliance with regulatory requirements to safeguard sensitive data.
• Security Awareness Training – Employees are often the first line of defense. nGuard provides training to help organizations recognize and avoid AI-related threats, including phishing and social engineering attacks.

DeepSeek’s security vulnerabilities, questionable data practices, and ties to Chinese state-affiliated entities present significant risks to users and organizations. Given the exposure of unencrypted data, leaked credentials, and potential government surveillance, businesses should reconsider using DeepSeek in any capacity. With increasing regulatory scrutiny and growing concerns from cybersecurity experts, the risks outweigh the benefits. Organizations should take immediate steps to secure their data and consult nGuard for tailored security solutions to mitigate AI-related threats.