Objave z oznako "Trustworthy AI"

Summary

The EU AI Act largely regulates the market placement of systems — risk classification, conformity, post-market surveillance of products. A…

Yesterday, I came across the recent OpenAI “wiki incident”, and it gave me a very different perspective on the governance challenges emerging with multi-agent AI.

It…

AI is moving beyond generation into consequential organisational use. It drafts, summarises, extracts, classifies, recommends and prepares material from which organisations create records, reports…

Generation is a capability. It should not always be the default execution mode.

Recently, I asked an AI system to compare two supplier quotations and prepare a price…

Artificial intelligence governance should not stop at principles, policies or model-level evaluation.

As AI systems become increasingly multi-model, agentic and embedded in…

Organizations are moving rapidly from AI experimentation toward decisions involving material investment, external providers, operating-model changes, workforce impact, governance obligations, and…

The EU Artificial Intelligence Act (Regulation (EU) 2024/1689) relies primarily on providers’ self-assessment to demonstrate compliance. Regulatory access under Articles 21 and 74 is conditional…

Introduction

Two metaphors keep surfacing when people try to imagine where unconstrained AI optimization could lead: collapse and transcendence. A black hole and a god are, in a sense,…