Articles - AIPMO (Page 3)

LLM Safety Benchmarking: What PMs Need to Know About Evaluating AI Model Safety

Your vendor says the model scored 94% on safety benchmarks. But what was actually tested — and what does that score obligate you to do? Here's how NIST MEASURE, the EU AI Act, and Singapore's Project Moonshot translate into PM accountability.

By AIPMO

/ 9 Mar 2026 · 11 min read

Open-Source AI: The Governance Challenges You Didn't See Coming

Open-source AI transfers governance responsibility to you. What your commercial vendor handled in the background — safety testing, documentation, incident response — now sits on your project plan. Here's what the EU AI Act and NIST say you're accountable for.

By AIPMO

/ 9 Mar 2026 · 10 min read

AI in Insurance: A PM's Guide to a High-Stakes Sector

Insurance AI is making decisions about who gets coverage, how much they pay, and whether claims are approved. EU AI Act Annex III classifies health and life insurance risk assessment as explicitly high-risk. Here's what NIST MEASURE 2.11, the NAIC Model Bulletin, and state regulators require of you.

By AIPMO

/ 29 Jan 2026 · 13 min read

Agentic AI: What Project Managers Need to Know

Traditional AI advises. Agentic AI acts — autonomously browsing, calling APIs, executing code, and delegating to other agents. The shift from output to action changes everything about how you govern AI projects. Here's what EU AI Act Article 14, NIST MANAGE 2.4, and Singapore's IMDA require of you.

By AIPMO

/ 29 Jan 2026 · 14 min read

Change Management for AI Projects: Preparing People for a New Way of Working

AI change management goes beyond training people on a new tool. It requires helping them develop new mental models for working with probabilistic systems — and addressing the trust, accountability, and expertise questions that conventional IT change management was never designed to handle.

By AIPMO

/ 29 Jan 2026 · 14 min read

Third-Party AI and Vendor Management: Risks You Don't Control

When you deploy third-party AI, you deploy the decisions its developer made about training data, fairness testing, and risk mitigation — and you remain accountable for the outcomes regardless of who built the system. Here's what PMs need to know before signing the contract.

By AIPMO

/ 29 Jan 2026 · 14 min read

Monitoring AI Systems in Production: The Work After Go-Live

AI systems degrade in production even when no one touches the code. Post-deployment monitoring is not optional: for high-risk AI it's a legal requirement, and for all AI it's the only way to know whether the system is still doing what it was designed to do.

By AIPMO

/ 29 Jan 2026 · 16 min read

Testing and Validation for AI Systems: More Than Accuracy

Every AI project needs testing — but the test plan looks nothing like traditional software QA. The NIST AI RMF TEVV framework requires testing across five dimensions: accuracy, fairness, robustness, safety, and explainability. A system can pass every conventional test and still discriminate.

By AIPMO

/ 29 Jan 2026 · 17 min read

Stakeholder Engagement for AI Projects: Beyond the Usual Suspects

AI systems affect people who never appear on a traditional stakeholder register — and under the EU AI Act, informing them and giving them a mechanism to contest decisions is a legal obligation. Here's how to build a stakeholder approach that goes beyond the usual suspects.

By AIPMO

/ 29 Jan 2026 · 14 min read

Human Oversight in AI Systems: Designing for Control

Human oversight of AI isn't a design preference — it's a legal requirement. EU AI Act Article 14 mandates five specific capabilities in every high-risk AI system. As PM, each one is a project deliverable, not a governance aspiration.

By AIPMO

/ 29 Jan 2026 · 35 min read

Model Cards and Datasheets: Documentation That Matters

AI projects need documentation that doesn't exist in conventional IT: model cards, datasheets, and system cards. Under the EU AI Act, technical documentation must exist before deployment — and be retained for 10 years. Here's what each document must contain and why.

By AIPMO

/ 29 Jan 2026 · 17 min read

AI Risk Registers: Beyond Traditional Risk Management

AI risk registers use the same mechanics as conventional ones — identify, assess, mitigate, monitor. What changes is the taxonomy of risks, how likelihood behaves at scale, and the fact that the register doesn't close at deployment. Here's what needs to extend.

By AIPMO

/ 29 Jan 2026 · 17 min read