Skip to main content
    Insights

    Analysis of AI governance, agent infrastructure and operating practice.

    Featured pillar: Sovereign AI, what it really means for the enterpriseLatest: iTmethods joins the Linux Foundation, FINOS, and the Agentic AI Foundation

    The New Stack

    Building AI-Native Organizations

    A blueprint for the infrastructure, teams, and operating model that separate companies building with AI from the ones being disrupted by it.

    AI Governance

    Securing the Agentic Era

    The security, compliance, and governance layer enterprises need before autonomous agents reach production.

    MCP Is Exploding. Your Governance Isn't Ready.OpenClaw: The Governance Failure We Saw ComingWhy 80% of Enterprises Are Stuck. And What the Control Plane Actually Looks LikeChamath Just Said What Every Enterprise CISO Already KnowsWhy Regulated Industries Hit a Wall. And What the New Stack Actually RequiresSecuring the Agentic Era: OpenClaw, NemoClaw, and the Governance Layer That's Still MissingAnthropic Just Made Every Desktop an AI Agent. Here's What That Means for Enterprise Governance.The Deadline Moved. The Evidence Problem Didn't.Why the AI Governance Stack You Have Was Built for the Wrong ProblemWhen Your Vendor's AI Ambitions Become Your Governance Problem96 Days, Maybe 600. Why the EU AI Act Delay Talk Doesn't Change What You Need to BuildAgent Sprawl Is the New Shadow IT. And the Numbers Are Worse Than You ThinkSR 26-2 Just Created a Governance Gap Banks Can't IgnoreInside OpenAI, the Platform Team Is the New Trust BoundaryAI Agent Operations Is the Third Pillar of Governed AIAnthropic Moved Twice. OpenAI Moved Twice. The Trust Boundary Moved With Them.15 Days. Six Vendor Moves. Three Breaches. One Threat Actor.Three Days. One Export Order. A Frontier Model Gone.Canada's Sovereign AI Stack Has One Layer Left to BuildThe New Standard for AI Trust Is Here. The Runtime Layer Is Not.The Five Questions Your SaaS Vendors Hope You Never AskThe Sovereign AI Trade Is Missing a Line ItemPalantir and NVIDIA Turned Control Into a Product CategoryThe Filing Is Not the WorkThe Answer Is an Estimate. Somebody Has to Check It.

    Latest

    CEO Insight
    AI Governance|Sep 9, 2026

    The Answer Is an Estimate. Somebody Has to Check It.

    At the G20 on 2 September, Alex Karp described the one job his own sovereignty stack hands to a log. Checking the work is a different job from holding the asset.

    Read Article

    Latest Articles

    AI Governance

    The Filing Is Not the Work

    Article 14 of the EU Cyber Resilience Act starts 11 September 2026. 24-hour early warning of actively exploited vulnerabilities. The industry will treat this as a filing problem.

    Sep 2, 2026
    AI Governance

    Palantir and NVIDIA Turned Control Into a Product Category

    An alliance of forty companies, a contract playbook, and a television appearance, all on the same Monday, four weeks after the partnership that set it up. Two of the largest companies in the industry just told several thousand enterprises that the control layer is theirs to hold, and put engineering behind the claim. They are right about the problem. Owning a layer is a commercial arrangement. Proving an action is an artifact.

    Jul 28, 2026
    AI Governance

    The Sovereign AI Trade Is Missing a Line Item

    In one week, sovereignty became a product category. Air-gapped. Never trained on. Does not log. Then the investor who named the trade said there is not a scintilla of evidence AI has lifted margins, and told CNBC that zero data retention does not cover what you click, because the industry does not know how to do that. Every one of these products moves the boundary. Not one produces the proof.

    Jul 14, 2026
    AI Governance

    The Five Questions Your SaaS Vendors Hope You Never Ask

    On live TV, Alex Karp handed enterprises the clearest AI vendor due-diligence framework of the year: five questions. He aimed them at the frontier labs. The worse answers sit down the SaaS stack, where nobody is asking and the terms change by default. Ask them in writing, per vendor, per tier.

    Jul 7, 2026
    The New Stack

    The Software Factory Is Going Dark. The Audit Trail Cannot.

    An autonomous AI software engineer just crossed 492 million dollars in revenue, with Goldman, Citi, and Santander as customers, and Gartner expects agents to write nearly half of enterprise code this year. The money is going to the workers. The durable advantage is the supervisor that governs the floor and proves what the agents did. That is the moat, and it is buildable now.

    Jul 1, 2026
    AI Governance

    The New Standard for AI Trust Is Here. The Runtime Layer Is Not.

    The Linux Foundation just launched the Appia Foundation with Google, Microsoft, OpenAI, Mastercard, and others, making verifiable proof of trustworthy AI an open standard. The piece still missing is continuous, agentic proof that holds when agents act and models change. That is the runtime layer, and it is buildable now.

    Jun 23, 2026
    AI Governance

    Canada's Sovereign AI Stack Has One Layer Left to Build

    Canada is building sovereign AI: a national strategy, billions in data centers, real champions on models, compute, and research. Every champion is on one side of the stack. The control and assurance layer that makes sovereignty operational and provable has no Canadian name on it. That is the gap, and the opportunity.

    Jun 17, 2026
    AI Governance

    Three Days. One Export Order. A Frontier Model Gone.

    On Friday a frontier model that hundreds of millions of people were using disappeared, withdrawn from the entire market by a US government export directive. The lesson is about dependency, not security: single-model dependency is a single point of failure, and the control layer outranks the model.

    Jun 15, 2026
    AI Governance

    Inside OpenAI, the Platform Team Is the New Trust Boundary

    OpenAI's own Data Infrastructure Engineering lead just described, in plain English, why agents need governance separate from the workload. She used the word 'adversarial.' She was describing OpenAI's own engineers' agents touching OpenAI's own platform.

    May 28, 2026
    AI Governance

    Anthropic Moved Twice. OpenAI Moved Twice. The Trust Boundary Moved With Them.

    A tale of two cities. SaaS wants your data. The regulated enterprise wants it back. In the same fortnight the vendors chose sides, Grafana, Mistral, OpenAI, and GitHub were breached by the same threat actor.

    May 26, 2026
    AI Governance

    15 Days. Six Vendor Moves. Three Breaches. One Threat Actor.

    OpenAI moved twice. Anthropic moved twice. Palantir, Dell. And in the same fortnight, TeamPCP breached Grafana, Mistral, OpenAI, and GitHub. The two companies that defined cloud-native AI in 2022 each made two distinct customer-side decisions in the same window.

    May 21, 2026
    AI Governance

    AI Agent Operations Is the Third Pillar of Governed AI

    Monitoring and assurance were built for the model era. The agent layer is where governance gets hard, and where most vendors stop. AI Agent Operations is the third pillar of governed AI, and continuous remediation is the loop-closing capability inside it.

    5 min read May 15, 2026Paul Goldman
    AI Governance

    SR 26-2 Just Created a Governance Gap Banks Can't Ignore

    The Federal Reserve just carved AI out of model risk management. The institutions that treat this as relief instead of a budget signal will be 12 to 18 months behind.

    4 min read May 12, 2026Paul Goldman
    AI Governance

    Agent Sprawl Is the New Shadow IT. And the Numbers Are Worse Than You Think

    97% of enterprises now run AI agents. Only 12% have centralized control. The 88-point gap between running and governing is the operational consequence of an executive perception gap. And the architecture answer is the same for AIGF v2.0, EU AI Act, DORA, and OSFI E-23 simultaneously.

    12 min read May 5, 2026Paul Goldman
    AI Governance

    96 Days, Maybe 600. Why the EU AI Act Delay Talk Doesn't Change What You Need to Build

    Brussels failed to agree on a 16-month delay today. Even if the delay eventually passes, regulated enterprises building governance infrastructure for the original August 2 deadline are still right.

    11 min read Apr 28, 2026Paul Goldman
    AI Governance

    When Your Vendor's AI Ambitions Become Your Governance Problem

    SaaS vendors are training AI on customer data by default. The FINOS AI Governance Framework, the EU AI Act, and DORA all have a position. Most enterprises have not caught up.

    10 min read Apr 23, 2026Paul Goldman
    AI Governance

    Why the AI Governance Stack You Have Was Built for the Wrong Problem

    Agentic AI is in production in regulated industries. The governance stack most institutions spent the last three years building was designed for a world that no longer exists. FINOS Open RegTech shows the architecture pattern that actually works.

    12 min read Apr 21, 2026Paul Goldman
    AI Governance

    The Deadline Moved. The Evidence Problem Didn't.

    The EU AI Act's high-risk deadline slipped eighteen months. If that changed your plan, your plan was never about evidence.

    8 min read Apr 14, 2026Paul Goldman
    The New Stack

    The Camunda 7 Fork That Financial Services Built. And Why It Changes Everything for AI Governance

    A FINOS community fork of Camunda 7 is now hosted under the Linux Foundation, with sponsors and contributors from Fidelity, Deutsche Bank, NatWest, Capital One, and BMO publicly visible on the project page. Here is why this changes the AI governance conversation entirely.

    12 min read Apr 6, 2026Paul Goldman
    The New Stack

    The Platform Engineering Pivot

    AI agents don't manage themselves. The teams that figure that out first will define what AI-native organizations actually look like.

    14 min read Mar 31, 2026Paul Goldman
    The New Stack

    Self-Hosted AI Agents Are Here. The Governance Isn't.

    Cursor's self-hosted cloud agents are a proof point for the AI-native infrastructure shift. And a warning about the governance gap that enterprises must close.

    12 min read Mar 26, 2026Paul Goldman
    AI Governance

    Anthropic Just Made Every Desktop an AI Agent. Here's What That Means for Enterprise Governance.

    Computer Use + Dispatch = autonomous agents on every MacBook. The governance conversation starts now.

    10 min read Mar 25, 2026Paul Goldman
    Bridge

    Securing the Agentic Era: OpenClaw, NemoClaw, and the Governance Layer That's Still Missing

    NemoClaw secures the container. Reign governs the enterprise. Here's why you need both.

    14 min read Mar 23, 2026Paul Goldman
    The New Stack

    Why Regulated Industries Hit a Wall. And What the New Stack Actually Requires

    Generic controls collapse when billions and lives are on the line. Here's what actually works in financial services, healthcare, life sciences, and defense.

    12 min read Mar 17, 2026Paul Goldman
    The New Stack

    Why 80% of Enterprises Are Stuck. And What the Control Plane Actually Looks Like

    The missing layer between AI-assisted and AI-native isn't a model. It's governance.

    12 min read Mar 11, 2026Paul Goldman
    The New Stack

    The AI-Native Stack: What It Actually Looks Like

    Three layers separate the companies building with AI from the companies being disrupted by it

    10 min read Mar 3, 2026Paul Goldman
    AI Governance

    MCP Is Exploding. Your Governance Isn't Ready.

    The five questions every organization should be able to answer about AI agent governance

    8 min read Feb 24, 2026Paul Goldman
    AI Governance

    Chamath Just Said What Every Enterprise CISO Already Knows

    Why data sovereignty, AI cost control, and attorney-client privilege demand governed infrastructure. Now

    8 min read Feb 14, 2026Paul Goldman
    Security Advisory

    OpenClaw: The Governance Failure We Saw Coming

    An autopsy of what happens when AI adoption outpaces control

    12 min read Feb 10, 2026Paul Goldman
    Continuous Agentic Assurance

    See governance in action.

    Book a twenty-minute walkthrough of the Reign platform.

    Talk to Us

    The Trust Layer.

    A weekly essay on governing agentic AI in regulated industries. No product pitches.