Amazon employees warn customers about DeepSeek’s privacy risks, promotes its own AI instead

Amazon

Amazon employees are warning customers about privacy concerns with DeepSeek, the popular Chinese AI model that gained success overnight. While Amazon initially rushed to add DeepSeek to its AI platform, Bedrock, a report by Business Insider revealed that the company is now changing its stance. Employees are being encouraged to warn customers about security risks and suggest Amazon’s own AI models instead.

DeepSeek gained attention in January for being powerful and affordable. Many companies wanted access, and Amazon moved fast to include it on Bedrock. Some employees found the approval unusually quick, but CEO Andy Jassy explained that Amazon was just meeting demand.

Now, Amazon is taking a step back. Internal documents accessed by BI reveal that employees are being asked to highlight DeepSeek’s privacy risks. The AI model collects user data, which could be stored on servers in China—something that worries many customers.

Along with privacy warnings, Amazon is promoting its own AI model, Nova. Employees are told to market that Nova is faster, more secure, and can handle more types of data, including images and videos. Unlike DeepSeek, Amazon claims Nova follows stricter “responsible AI” standards.

Internally, Amazon is discouraging employees from using DeepSeek at work, the report says. Employees receive warnings not to share sensitive information with the model, similar to existing restrictions on ChatGPT.

Even though Amazon still offers DeepSeek on Bedrock, it is also working on its own advanced AI to compete with it. The company is making sure customers know the risks—and guiding them toward its own solutions.

AI is changing fast, and competition is heating up. But Amazon wants customers to think twice before trusting DeepSeek.