Stephen Nyarko

AI governance specialist. Creator of the S.A.F.E.™ & V.E.R.I.F.Y.™ frameworks for safe AI use.

Add social links

Your AI assistant can be factually accurate and still give the wrong answer.

Aug 15, 2026

•

8 min read

Your AI assistant can be factually accurate and still give the wrong answer.

The model did not hallucinate. It retrieved accurately, from a document your firm should have retired six months ago.

Stephen Nyarko

Your Firm Approved One AI System. It Connected to Seventeen.

Aug 11, 2026

•

4 min read

Your Firm Approved One AI System. It Connected to Seventeen.

AI registers list models. The governance question attaches to connections, and almost nobody is counting those.

Stephen Nyarko

The Advice Boundary Has Three Positions Now. Most Controls Only Know Two.

Aug 11, 2026

•

6 min read

The Advice Boundary Has Three Positions Now. Most Controls Only Know Two.

Targeted support came into force on 6 April. It is a rule change that permits something rather than restricting it, which is exactly why it breaks controls quietly instead of loudly.

Stephen Nyarko

The EU AI Act Moved Its Deadlines Three Weeks Ago. Most Published Guidance Still Has the Old Ones.

Aug 11, 2026

•

5 min read

The EU AI Act Moved Its Deadlines Three Weeks Ago. Most Published Guidance Still Has the Old Ones.

The Act was amended on 27 July. The high-risk timetable shifted by sixteen months, two new prohibitions appeared, and the change is recent enough that most material a firm is working from predates it.

Stephen Nyarko

Three things about AI regulation that are actually current

Aug 11, 2026

•

5 min read

Three things about AI regulation that are actually current

The AI Act dates that moved, the third position on the advice boundary, and what your agents can actually reach.

Stephen Nyarko

Three AI rules that are actually yours this week

Jul 27, 2026

•

4 min read

Three AI rules that are actually yours this week

The AI Act deadline that didn't move, the FCA's real message, and the agent authority gap.

Stephen Nyarko

Your AI Agents Have More Authority Than Your Governance Does

Jul 25, 2026

•

5 min read

Your AI Agents Have More Authority Than Your Governance Does

Enterprises are handing AI agents the power to act, not just advise, faster than they are building the controls to govern them. The failure mode is not a dramatic error. It is that authority creeps upward until no one can say who made the decision.

Stephen Nyarko

The FCA Just Said It Won't Regulate AI (For Now). Here's What That Actually Means.

Jul 25, 2026

•

6 min read

The FCA Just Said It Won't Regulate AI (For Now). Here's What That Actually Means.

On 6 July 2026 the FCA published its most significant statement yet on AI in financial services. The headline is that nothing changes. The reality is that the accountability was always yours, and the regulator has just confirmed it is staying that way.

Stephen Nyarko

Jul 24, 2026

•

9 min read

The AI Act Deadline That Didn't Move

Most coverage said the EU delayed the AI Act. Part of it did move. The part that applies to your chatbot, your marketing images and your client-facing content did not, and it starts on 2 August 2026.

Stephen Nyarko

Regulatory Technology