PUBLIC AI

The Gap Between What Governments Claim and What Actually Happens

By: Eduardo Valencia

Narrated by: Virtual Voice

You've sat through the slide deck. The commitment was evident.

Then you returned to your floor — where a policy analyst had been using AI for six months without authorization, and a data team had integrated a third-party AI service transmitting information that should never have left the organization.

None of it was malicious. None of it was covered by the strategy.

PUBLIC AI is written for the senior civil servants who live in this gap — secretaries general, deputy directors, heads of unit — the layer where ministerial commitments become operational behavior, and where the distance between the two is felt most acutely.

Eduardo Valencia identifies the three governance failures that produce the gap: the national AI strategy that generates commitment without operational guidance; the shadow AI that populates every organization whose employees are more capable than its tools; and the institutional fiction that allows both to coexist without accountability.

What you will find:

• 
  
• Why 30–60% of public sector employees use AI tools without authorization — and why the first response should never be prohibition
• 
  
• What the EU AI Act actually requires from public administrations in practice
• 
  
• The Shadow AI Audit methodology: visibility before strategy
• 
  
• A twelve-week plan for governance that functions rather than governance that documents
• 
  

This is not a policy book. It is a management book for the person who must convert political commitment into organizational behavior.

Second volume in the CIVIC AI Series.