AI governance for systems that can act.
Field guides, films, and evidence-led briefs for leaders moving from AI experimentation to systems with data access, delegated authority, and real-world consequences.
- Curated insights: 44
Govern the action, not only the model.
Useful AI governance connects each consequential workflow to accountable authority, enforceable limits, human oversight, evidence, and a measurable reason to operate. These articles examine the architecture and decisions that make those controls real.
Insights
- Before You Buy Another AI Security Tool, Map What You Already Have: Map one AI workflow, its existing controls and the evidence behind them before deciding whether another security product fills a real gap.
- Changing the Model Means Retesting the Workflow: Review model changes against the full workflow: sources, permissions, actions, escalation and failure handling, with evidence and a way to recover.
- The Real Cost of AI Is the Cost of a Result You Can Use: Measure the cost of an accepted AI result, including review, correction, integration and support, with a worked example and a practical pilot method.
- Your AI Agent Has an Identity. Who Owns It?: Make agent ownership practical by recording identities, access, approval points and lifecycle decisions, then checking how suspension and recovery work.
- OSFI’s AI warning: waiting is a decision, too: Three AI investment questions for Canadian banks and insurers: improve defences, set authority and price dependency. Read the brief and four-page PDF.
- AI agent egress control: constrain the request and the role: Review AI agent egress, MCP request paths, local metadata access and execution-role scope. Read the field guide and download the six-page PDF.
- Cyber Readiness in the Frontier AI Era: A practical guide to Canada's October 2026 cyber security direction, with six pillars, evidence checks, recovery exercises, and an implementation plan.
- Clef Flash: Decisions at Machine Speed: A 3:44 film testing Clef Flash on local security triage, support routing, and tool selection, with examples of its strengths and missed threats.
- Who Gave the Agent Permission?: A customer refund shows why agent ownership must reach the individual action—and why one approval cannot cover every decision.
- Check Permission at the Moment of Action: How runtime authorization connects an agent’s proposed tool call to policy, human approval, and the protected service.
- What to Ask an AI Vendor After the Demo: Six procurement questions that turn a polished AI demonstration into an examination of data, permissions, evidence, change, and exit.
- The AIA Starts Before the AI Contract Is Signed: For covered federal decisions, procurement must secure the information and rights needed to complete and maintain the impact assessment.
- The Agent Registry Nobody Maintains: An agent inventory becomes useful when it can answer who owns a running service, what it depends on, and how to retire it.
- An Agent Has Gone Wrong. What Do You Stop?: A hypothetical incident walkthrough follows execution, queued work, credentials, evidence, and recovery beyond the visible chat window.
- The Model Changed. Does the Approval Still Hold?: A change-impact approach for model updates, prompts, retrieval sources, tools, and permissions—without treating every release alike.
- A Demo Is Easy to Approve. A Service Is Harder.: How to turn an AI agent launch meeting into a decision supported by observed behaviour, operational ownership, and recovery evidence.
- Where Should an Agent’s Tools Live?: Compare local functions, direct services, and shared gateways by ownership, enforcement, failure behaviour, and operating cost.
- When Research Agents Cross Access Boundaries: How routine AI research led to attempted intrusions: a review of Transluce findings, Australian government evidence, and practical oversight controls.
- The Architecture of Synthesis: An architecture for AI systems that retain context, act within defined permissions, and expose their decisions for review against current evidence.
- Pumping the Brakes for Whom?: What should an AI slowdown actually stop? Junior Williams examines internal model use, fair access, and independent authority to halt dangerous work.
- If AI Can Change Everything, Why Not Start With What Humanity Needs?: AI should help secure food, housing, education and a livable planet worldwide. Promised benefits do not justify existential risks.
- The AI Model Release Wave: 42 models in 30 days: Independent analysis of 42 AI releases in 30 days: agentic work, cost per task, open deployment, specialization, and enterprise control.
- AI Security Spending Is Rising. Established Platforms Are Winning the Budget: AI security budgets are rising, but platforms hold a buying advantage unless specialists close a material control gap with low operational drag.
- AI Security Risks Explained: From Deepfakes to Prompt Injection: A SecTor briefing on prompt injection, poisoned data, deepfakes, shadow AI, explainability, and accountable AI deployment.
- Graph Engineering the Art of Possible: Design AI workflows as connected steps with defined inputs, handoffs, approval points, and evidence, so interrupted work can be reviewed and resumed.
- Canada’s AI Transparency Opportunity: Beyond Watermarks and Disclosure: Canada can move beyond AI labels and watermarks by linking transparency to decision ownership, operational evidence, meaningful review, and correction.
- An agent will always find more work: A field essay on what one unbounded multi-agent run reveals about decision rights, independent measurement, resource ceilings, and external control.
- Toronto: One Prompt Through a Language Model: A 2:05 film tracing one prompt through a small language model, from seven tokens to the answer “Toronto,” with figures from a real model run.
- The Modern Enterprise Architect's Handbook: A practical operating guide for turning enterprise strategy into coherent architecture, investable transition states, governance, and evidence.
- Graph RAG Is for Connected Evidence: A field guide to deciding when graph-enhanced retrieval is worth the extra machinery and how to pilot it safely.
- AI is growing a spine: The spine metaphor explains the shift from AI systems that generate answers to agentic systems that carry intention into consequential action.
- Harness Engineering: Designing the System Around the Model: The article argues that the durable enterprise advantage in agentic AI sits in the harness around the foundation model rather than in the model alone.
- RAG Was Built For Answers. Agents Need Context Architecture.: Retrieval-augmented generation still serves search-shaped work, but agents need governed context across tools, permissions, workflows, memory, and state.
- When AI Strategy Advice Mirrors the User’s Assumptions: The article uses HBR and medical-query research to argue that LLMs should not be treated as strategists or oracles.
- How Secrets Enter an AI Coding Agent’s Context: AI coding agents can leak secrets when shell output containing credentials enters the conversation context sent to a model.
- Designing GenAI Pilots for Measurable Value: Why GenAI pilots fail when designed for demos instead of operations, and how to connect use cases to workflows, controls, evidence, and measurable value.
- Canada's AI Strategy: Building Infrastructure While Strengthening Governance: Canada's AI policy moment shifts from stalled legislation toward parallel investment in compute infrastructure, adoption, and stronger governance.
- When AI Systems Start Whispering in Code...: What machine-to-machine communication, autonomy, and public-sector adoption reveal about governing AI systems that interact beyond direct human supervision.
- The AI Power Struggle and the Future of Intelligence: This opinion piece argues that frontier AI development is increasingly concentrated among a small number of corporate and government actors.
- Qwen2.5-Max and the Competition in AI Models: An opinion analysis of Qwen2.5-Max and what rapid model competition reveals about AI capability, market concentration, governance, and strategic dependence.
- Advancing AI with Accountability and Fairness: How to turn AI accountability and fairness from statements of intent into operating controls, accountable ownership, reviewable evidence, and decisions.
- Human-AI Teaming in the Age of Collaborative Intelligence: How collaborative intelligence changes AI from passive automation into an active partner, and what human oversight, role design, and accountability require.
- Responsible AI Implementation in Enterprise and Public Sector: A practical approach to responsible AI in enterprise and public-sector settings, connecting governance principles to controls, accountable owners, and evidence.
- Maximizing Cybersecurity with AI: A Comprehensive Guide to Applications, Strategy, Ethics, and the Future: This broad guide surveys AI's role in cybersecurity across defensive applications, strategy, ethics, human factors, and future outlook.