Filtered by: AI & Agents × clear
SR 26-2 punted generative AI out of model risk scope and told you to govern it anyway. The FINOS AI Governance Framework is a free catalogue of 46 risks and controls, with the regulatory mapping already done.
I was challenged to work out what it would take to think of agents as colleagues. Taken seriously it becomes an operating model: identity, scoped access, governed onboarding context, supervision, evaluation, deprovisioning. Here's what worked when I built that, and the three places it broke.
The new model risk guidance excludes generative AI in a single footnote and defers the rest to a future rulemaking. But out of scope isn't unregulated - a stack of existing law still governs every AI decision you ship.
Every ontology article leaves you more confused than when you started. Here is the opposite - one real ontology running inside a banking AI, every piece of jargon translated (taxonomy included), the drift it killed, and what it becomes at enterprise scale.
Self-hosting an open model is not what keeps your data safe in a regulated industry - owning the front door is. I built an AI governance gateway on GCP with PII screening, token budgets, and a full audit trail for under 10 dollars a month, and it changed my answer on the whole build-vs-buy question.
I kept three copies of what my analyst agent was allowed to touch, and they drifted. Here's how the Open Knowledge Format cut it to one git-versioned source, compiled once and guarded by CI.
Ten quick questions on building retrieval-augmented systems that scale under load and don't leak data they shouldn't — access control, prompt injection, PII, evals, and caching. No reading required, just test yourself.
My chat bot confidently lied about my own writing. The bug wasn't retrieval - it was an ambiguous sentence in the source and a stale index. Why RAG is a data problem before it's a retrieval problem.
