The launch of a lower-cost method for creating generative AI models with minimal guardrails has sparked controversy. This has highlighted the need for governance policies, which managed service providers (MSPs) will often enforce. Developed by researchers in China, the DeepSeek AI model uses techniques that require less memory and rely more on machines rather than humans to provide reinforcement training.
DeepSeek AI: Controversy, licensing restrictions, and data privacy concerns
DeepSeek is available under an open-source license, making it attractive to many end users who are already taking advantage of downloading it directly or simply accessing it via mobile applications that can be found in the online stores managed by Apple or Google.
However, there are DeepSeek AI licensing restrictions that can result in data finding its way back to data centers in China. These include monitoring interactions and usage across your devices, as well as analyzing how people are using it. Additionally, organizations can use data to train and improve models while complying with any legal obligations. DeepSeek AI reserves the right to share collected information with its advertising and analytics partners. This includes sharing data with third parties during corporate transactions, such as mergers, asset sales, or changes in company control. Those terms of usage align with the laws of the People’s Republic of China (PRC).
DeepSeek AI also broadly disclaims any liabilities in connection with its services. There is little data on how training data was or the guardrails that might have been set up, and there are already reports detailing how sensitive DeepSeek AI models are in exposing data.
Finally, DeepSeek AI requires organizations to clearly indicate when they generate output using AI. They must also proactively verify the authenticity and accuracy of the output.
MSPs play a key role in enforcing policies
Not surprisingly, some organizations and even some countries have already moved to ban the usage of DeepSeek AI, even as others excitedly experiment with a model that can now be found on cloud services. Of course, banning the usage of something and enforcing that policy in the age of shadow IT are not one and the same. Many organizations want to ban DeepSeek AI. They will need the expertise of their MSPs to enforce the ban on end users.
Of course, DeepSeek AI is only the tip of an AI governance and security challenge that every organization must navigate. Many AI-savvy MSPs had already been having this conversation with their customers even before the rise of DeepSeek AI. DeepSeek AI has generated significant attention in recent weeks. As a result, many more business and IT leaders are now open to having the discussion. This shift highlights the growing urgency around AI governance and security.
Photo: jittawit / Shutterstock
This post originally appeared on Smarter MSP.