DeepSeek, a new AI tool, is not currently an approved AI tool for campus use, but UAB experts are offering tips for those who want to try it on their own.

UAB has several AI tools for campus use that have been vetted heavily by professionals in the UAB AI Advisory Committee, AI testing groups and UAB IT’s Information Security team. Campus users who wish to use the approved AI tools can find them on the official AI website.

“Our team spends a great deal of time vetting these tools, to help upkeep a data ecosystem that is stable and overall secure. Where our data is going and how it is being used is a huge priority,” said Rob Ferrill, assistant vice president and chief information security officer.

For those curious about emerging AI models, there are safe and unsafe ways to access DeepSeek. As with any AI tool, users should follow UAB’s AI principles and guidelines.

Safe:

• Chat services, similar to ChatGPT and Microsoft Copilot, may offer DeepSeek as a search option. Users should only enter public data into the model if using it.
• Several programming options are available, through Amazon Web Service’s Bedrock or local open-source models. To get approval for use, submit a request to UAB IT Research Computing.

Unsafe:

• Web access. The DeepSeek website reportedly has a vulnerability, where significant amounts of data were exposed for bad actors to gain sensitive information. Further, the Terms and Conditions legal agreement allows the vendor to provide all data to China.
• Mobile access. The app for this AI tool requests access to more data than appropriate for a mobile app.
• DeepSeek API. This is targeted for programmers and is not approved for campus use.

AI tools currently approved for use on campus follow UAB’s data classification guidelines.  This ensures that UAB data is protected, and it is important that users of AI also follow these guidelines.