All of the following issues are unique for proprietary AI model deployments EXCEPT?

Biasis a common risk acrossboth proprietary and open-source models, andnot uniqueto proprietary deployments. All AI systems — regardless of origin — require evaluation for fairness, accuracy, and representativeness.

From theAI Governance in Practice Report 2024:

“Bias, discrimination and fairness challenges are present in both open and closed models, regardless of how the model is sourced.” (p. 41)