It seems like it’s all AI, all the time recently in Washington D.C. A flurry of executive orders and the Trump Administration’s AI Action Plan were announced last week. Coverage I’ve read explains that the plan focuses on encouraging innovation, speeding infrastructure development such as data centers, and implementing export controls on technology such as semiconductors. It differs significantly from the more cautious approach to AI by the Biden Administration.
The latest headlines reminded me I hadn’t posted a presentation I gave to my GovCon group on the first AI Executive Order, EO 14179, issued in January. Fellow GovCon Ideator Kathy Swacina presented with me on 4/29. We take a deeper dive into the EO language, highlight the differences from past EOs on AI, and suggest a possible model for AI development. An effective option is the TAIMScore model from the Holistic Information Security Practioner Institute (HISPI).
I reached out to a couple of IT clients for their views of EO 14179:
“I’ll first take the glass is half full attitude on this memo. it addresses a lot of the needed “how” that was ambiguous before as far and processes and procedures. It specifically calls out driving the
departments mission vs just efficiencies in the department. it will give teams a lot of opportunity to move forward with POC’s.”
“On the glass is almost empty side of the equation. There is no significant change to actual level of risk tolerance, governance and mitigation that must be performed. There is still plenty of
opportunity for a risk adverse Sr Manager at a department to say no.
As before the adoption of of AI will come down to agencies with leadership and actual desire to solve problems with AI.”
Here are the questions left unanswered in EO 14179, and which still are after the recent Action Plan:
- How do we safeguard industry proprietary information?
- How do we protect American citizen PII?
- How do we ensure the industry partners are following safe, ethical, trustworthy and transparent development processes?
- How do we close the gap between federal and state AI regulatory regimes?
You can download the full presentation here.