This piece reflects the opinions of its author. We at GAINS remain enthusiastic about the responsible use of AI in higher education. However, we believe that Pitt’s community should have access to relevant information when deciding when and how to use AI, and we publish this Bite as part of our ongoing commitment to open discourse and dialogue.
By Rowan James
27 May 2026
Between October 2025 and January 2026, Pitt announced and slowly unveiled its partnership with Claude for Education, an Anthropic AI service, as well as Amazon Bedrock powered by Amazon Web Services, for integration into PittGPT. PittGPT, Pitt’s generative AI services, are intended for the creation of an “AI-powered campus of the future,” without any clear detail as to the intentions or directions of this project. While Pitt continues to push for an innovation-forward approach to AI, Pitt’s student body, faculty, and staff are growing more and more concerned about the immediate implementation of generative artificial intelligence into the very mechanisms of the campus and the University.
To understand the broader implications of this contract, we must look at the backgrounds of the other companies involved in this. Beginning in 2024, Anthropic partnered with data-mining institution Palantir to give the United States government access to AI models for intelligence, defense operations, and services. Much like Palantir’s intentionally-elusive public image, the nature of the PittGPT contract remains hidden to students and faculty alike. Pitt’s own services portal assures the Pitt community that “PittGPT operates within a closed system network, ensuring that no user data is used for retraining or shared externally.” However, a January 2025 email to the student body noted exceptions for “rare circumstances…as permitted under applicable University policies and federal, state and local laws” without any further explanation of what these circumstances consisted of, nor which University policies would be considered “applicable.” Locations in which data may be stored are not publicized, nor are security packages that the application has passed or is in the process of passing.
Pitt’s elusive explanations surrounding the collection and storage of data from its AI services not only mimic Palantir, but also showcase the University’s willingness to normalize ties. The implications of Palantir normalization are immediate and dire: last year, Palantir was paid 30 million dollars by United States Immigration and Customs Enforcement to build ImmigrationOS, a surveillance platform that will pull from vast amounts of data and information to detect immigrants and non-citizens who allegedly pose potential “threats” to national security. Given the current ICE escalations all across the country, including here in Pittsburgh, as well as the fact that Pitt’s communications mention retraining but are unclear about other factors, it is not at all unreasonable to believe that PittGPT may pose a larger security threat to the University’s immigrant community than the University would like to let on.
The safety concerns of international students are one of many reasons voiced to me that Pitt students, faculty, and staff alike are all concerned about the implications of this contract. For a technology such as PittGPT to be used, Pitt must be prepared to publicly answer the following questions among others:
How are international students and faculty being protected? Where are the lines drawn as far as which “University policies” can override a student’s right to data privacy? Are you willing to disclose how much was paid for this contract? How are faculty and staff consenting to the use of this technology? How can you bring yourself to justify the normalization of ties with companies such as Palantir?
If these questions cannot be answered publicly and in good conscience, the University must consider divestment from the artificial intelligence machine in all its forms.