Pelz-Sharpe said that at best, government departments might over time adopt some of the blueprint’s recommendations and provide some transparency to their use of AI systems. But he said it’s a stretch to expect public tech companies to follow the blueprint voluntarily.