The Invisible Gap: What Article 26(5) Actually Deems Fulfilled | Governed by Design #9

The Invisible Gap: What Article 26(5) Actually Deems Fulfilled | Governed by Design #9

There's a sentence in Article 26(5) of the EU AI Act that reads like a gift to financial institutions: the monitoring obligation "shall be deemed to be fulfilled" by complying with the internal governance rules you already follow. Most institutions read it once, file it under relief, and move on.

In this episode, Alex and Jaime give it the second reading, the one where the work starts. The carve-out is a bridge between two rulebooks, and a bridge carries load in both directions. You'll learn:

  • What "deemed fulfilled" actually routes, and the three things it does not do
  • Why the simplification is operational, not substantive: the burden moves from building another framework to proving the existing one reaches the agent
  • The scope clause most first readers skip: who the carve-out covers, and who it doesn't
  • Models vs. agents: why machinery built for periodic validation struggles with reasoning drift, autonomous tool use, and loops that close without a human
  • A walk across the bridge: one agentic fraud-detection system, two readings, two very different outcomes
  • The DORA span: one event, potentially two regimes, one incident pipeline, and the stacked invisible gaps if the agent is mapped into neither
  • The Perimeter Test: six questions, one honest hour, every "no" pre-formatted as a Gap Register entry
  • The announcement: the six primitives are becoming a book, and this episode is drawn from its second chapter

Key Insight: The financial institution doesn't get a simpler AI Act. It gets an AI Act that assumes its existing governance is already excellent, and then holds it to that assumption. Nobody decided to leave the agent outside the governance perimeter.

The perimeter was drawn around models before the agents arrived. The gap isn't in your policy, it's in your scope.

Resources mentioned in this episode:📄 Full extended edition + downloadable Perimeter Test (six questions, one page, free): themohamedadam.substack.com📰 LinkedIn newsletter version: Search "Automate & Elevate" on LinkedIn🔗 Generate all six governance primitives, free: workspace.aistreamlinehub.com

Det här avsnittet är hämtat från ett öppet RSS-flöde och publiceras inte av Podme. Det kan innehålla reklam.

Avsnitt(39)

The Rehearsal: Proving You Can Stop the Agent Before You Have To | Governed by Design #8

The Rehearsal: Proving You Can Stop the Agent Before You Have To | Governed by Design #8

You filled in the Suspension Authority Ledger. A name, measurable triggers, an ordered notification chain. One question remains: is any of it true?In this episode, Alex and Jaime take Edition 7's sixt...

10 Juli 19min

The Authority to Stop: Who Can Halt the Agent When It Goes Wrong | Governed by Design #7

The Authority to Stop: Who Can Halt the Agent When It Goes Wrong | Governed by Design #7

An AI agent is in production and starts doing the one thing it was built never to do. Who has the authority to stop it, right now, this afternoon?We closed this series at five primitives. This episode...

1 Juli 20min

The Implementation Path: From Framework to Reality | Governed by Design #6 (Series Finale)

The Implementation Path: From Framework to Reality | Governed by Design #6 (Series Finale)

After five editions building a vocabulary for AI agent governance, the question that kept coming back was the same one. Where do we actually start?In this final edition of Governed by Design, hosts Al...

20 Maj 24min

Audit Trails That Survive Scrutiny | Governed by Design #5

Audit Trails That Survive Scrutiny | Governed by Design #5

Most agent systems have logs. Few have audit trails. Almost none have evidence architecture, the version that holds up under regulators. This episode introduces the Handoff Receipt: the unit of eviden...

12 Maj 24min

Who's Accountable When the Agent Acts? | Governed by Design #4

Who's Accountable When the Agent Acts? | Governed by Design #4

An agent does something unexpected. You know what happened. You don't know who's responsible.In multi-agent systems, accountability doesn't just get complicated, it disappears into the handoff. In thi...

5 Maj 19min

Human Oversight That Doesn't Become a Bottleneck | Governed by Design #3

Human Oversight That Doesn't Become a Bottleneck | Governed by Design #3

Most teams say they have human oversight. What they often have is a human approval queue.In this episode, hosts Alex and Jaime break down the Oversight Spectrum, a framework for designing human oversi...

28 Apr 25min

Governing AI Agents: The Decision Boundary Contract

Governing AI Agents: The Decision Boundary Contract

"The agent can process refunds."That's not a boundary. That's a capability statement.A boundary tells the agent what it's allowed to do, and what to do when it's not sure.In this episode of Governed b...

14 Apr 20min

Populärt inom Teknik

uppgang-och-fall
elbilsveckan
market-makers
rss-uppgang-och-fall
skogsforum-podcast
natets-morka-sida
developers-mer-an-bara-kod
rss-elektrikerpodden
allt-du-behover-veta-om-ny-teknik
bli-saker-podden
rss-en-ai-till-kaffet
rss-veckans-ai
rss-technokratin
rss-podden-om-dronare
rss-heja-framtiden
rss-laddstationen-med-elbilen-i-sverige
rss-inet-patch-notes
snacka-bil-med
hej-bruksbil
rss-upplyst-entreprenordirektor