娇色导航

Skip to main content

Restricting use of DeepSeek at UVic

May 01, 2025

DeepSeek is a Chinese artificial intelligence (AI) company that specializes in open-source large language models. It recently launched a free AI-powered chatbot which has gained significant attention.

Risks

Independent experts have reported serious privacy and security concerns with DeepSeek’s applications. BC provincial and federal governments have banned DeepSeek on their devices and networks. Other universities are taking measures to ban, restrict, or strongly caution against using DeepSeek.

The specific risks of DeepSeek use are:

  • exposure of personal and sensitive data
  • increased risk of surveillance
  • extensive data collection and sharing
  • risk of credential theft
  • regulatory and compliance risks

Response

UVic has reviewed expert findings and has determined the following:

  1. DeepSeek web and mobile applications accessed via web browser or mobile application do not meet UVic’s information security standards.
  2. UVic will restrict access to DeepSeek applications on the UVic network and for UVic-owned devices. This includes all users and devices on-campus, on UVic wireless networks, or using UVic VPN.
  3. UVic will restrict the use of DeepSeek on UVic-owned or personal devices used for university business, including laptops and phones. UVic also discourages the use or installation of DeepSeek on personal devices.

UVic understands the use of AI tools can provide benefits for learning and research, and has published a and Guidelines for Using Generative AI in Research. UVic users can use Microsoft 365 Copilot Chat as an approved Generative AI solution. For those who require it for research purposes, the DeepSeek Model (currently not known to be high-risk) can be downloaded and run locally.

Additional information

Information Security Standards are published under the authority of UVic Information Security Policy IM7800. Users of UVic’s information systems have a shared responsibility for information security and are expected to comply with Information Security and Acceptable Use policies, including security standards and guidance provided in this notice.