Eiropas profesionāļi sadarbojas modernā birojā

Apskats

Īss, Eiropai veltīts apskats par MI juridiskajā praksē. Pierādījumu līmeņi, jurisdikcijas konteksts un avotu saites katrā izdevumā.

Ieteicamais

Izdevums #33

When the Models Get Better, Does the Problem Go Away?

2026. gada 25. aprīlis

Sullivan & Cromwell — the firm that advises OpenAI on the safe deployment of AI — just filed a Chapter 15 emergency motion riddled with AI hallucinations. Liga reads the apology letter and asks: if better models still produce undetectable errors, what is actually broken? Four bets companies are making, one structural question, and why the simulator is still missing.

Lasīt izdevumu

2026. gada 25. aprīlis

When the Models Get Better, Does the Problem Go Away?

2026. gada 15. aprīlis

Why We Added a Seventh Pillar: AI Decision Boundaries

We added a seventh pillar to the Twin Ladder Standard — AI Decision Boundaries — because the gap between policy and execution kept showing up in our assessments. ING and ABN Amro automating AML decisions; vendors pre-filtering choices before humans see them. The other six pillars do not catch this. The seventh does.

2026. gada 29. marts

The Governance You Didn't Delegate: When Your SaaS Vendor Makes Decisions Before You See Them

Most organisations believe they have AI governance because they have policies, training, and monitoring. But none of those things can answer the question that actually matters in enforcement: when your AI system executed that action — was it allowed to? We introduce three governance models, expose the vendor pre-filtering problem, and ask whether opaque SaaS systems are structurally non-compliant under Article 4.

2026. gada 28. marts

Marketing's AI Compliance Blind Spot: When the Heaviest AI Adopter Can't Verify What It Publishes

Marketing teams account for 35% of enterprise generative AI spend — more than any other function. Yet fabricated statistics, hallucinated sources, and unverified regulatory claims are reaching audiences daily. We examine the layered exposure across GDPR, DSA, and the AI Act.

2026. gada 26. marts

GDPR vs the AI Act: Why Your Compliance Playbook Will Not Work This Time

Three times last week I was asked whether the GDPR programme 'covers' the AI Act. It does not. We examine four fundamental distinctions between guardrail regulation and process transformation — and why Article 4, binding since February 2025, demands competence that cannot be centralised.

2026. gada 21. marts

HR Is Ground Zero for Article 4: Why People Teams Face the Highest AI Compliance Risk

HR teams deploy more high-risk AI systems than any other function — from CV screening to performance prediction. Yet most recruiters cannot explain how their AI shortlisted one candidate over another. We map the exposure across Annex III categories and five European enforcement mechanisms already in motion.

2026. gada 14. marts

When AI Training Isn't Training: The Gap Between What Companies Buy and What Article 4 Requires

The EUR 2.4 billion European AI training market is selling certificates, not competence. We review 31 programmes and find that 87% allocate less than 10% of time to AI limitations and failure modes. Five markers distinguish genuine competence-building from checkbox compliance.

2026. gada 14. marts

The Procurement Blind Spot: When Buying AI Means Buying Compliance Risk

Most companies evaluate AI vendors using traditional software procurement frameworks. But Article 4 places the literacy obligation on the deployer — the buyer — not the seller. We introduce five RFP questions every procurement team should ask and a Vendor Competence Framework for scoring AI suppliers.

Apskatu arhīvs