DL 432

Normal Conditions

Published: May 13, 2026 • 📧 Newsletter

We've spent years describing a gap between where the technology is and where the governance, accountability, and institutional structures are.

This week, the gap stopped being a warning and started being the operating condition.

Three failures ran simultaneously. Technical infrastructure, epistemic infrastructure, and the accountability systems meant to catch both. They didn't wait for each other.

A note: Substack is now the delivery engine for this newsletter, but the broader archive and connected notes still live at digitallyliterate.net. The slower reflections continue at wiobyrne.com.

🏫 When Institutions Become Fragile Infrastructure

Two things failed in education this week. One was loud. The other barely registered.

The Canvas breach disrupted finals for thousands of students. A ransomware group exploited a vulnerability, Instructure negotiated with the ShinyHunters group to secure stolen data, and universities scrambled mid-semester with no real backup because Canvas isn't a tool that sits alongside the institution anymore. In many places, it has effectively become the institution.

This is a predictable outcome of K-12 and higher ed's decade-long reliance on SaaS (Software as a service). Software-as-a-Service means you rent the software, and it lives on someone else's computers. You just log in through a browser instead of owning or installing anything yourself. We spent years asking whether digital systems made learning more efficient. We spent almost no time asking what happens when they fail.

⚖️ The Courts Are Becoming AI Regulators

The AI industry had another week where the legal system appeared to move faster than formal regulation.

The Musk/OpenAI trial generated headlines across nearly every major technology publication. Testimony included accusations of manipulation, discussions of enormous financial stakes, and repeated questions about governance inside companies now shaping global infrastructure.

At the same time, courts and attorneys general are increasingly becoming the primary mechanism for AI oversight:

• Parents are suing OpenAI after ChatGPT allegedly encouraged a deadly drug combination to a teenager.
• Major publishers are suing Meta over claims that Llama models reproduced copyrighted material "word for word."
• A judge ruled that government agencies using ChatGPT to evaluate "DEI" compliance violated basic legal process protections.

Taken separately, these look like disconnected legal disputes. Together, they point toward something larger.

The deployment of AI systems is moving faster than democratic governance structures can respond, so accountability is increasingly being negotiated through lawsuits after harm has already occurred.

Technology companies often describe regulation as the force that slows innovation. But in practice, the absence of governance rarely removes friction. It simply transfers the burden downstream onto courts, schools, families, workers, and the public.

🔌 AI Is Rewriting the Physical World

The part of this story that gets the least coverage is what's happening beneath the interface.

Not a story about apps and prompts. Instead, a focus on electricity, water, supply chains, and labor.

A data center consumed 30 million gallons of water undetected before anyone raised a public flag. The largest U.S. power grids are straining under AI-related demand. Denmark's clean-energy infrastructure is being overloaded by data center expansion. These are more examples of what the buildout actually looks like, away from the product announcements.

Yet, with all this buildout and buzz, it's interesting to connect it to labor signals. Cloudflare had its best quarter and cut 1,100 jobs, attributing the reductions to AI agents. GM fired 600 IT workers to hire AI engineers. GitLab described headcount reductions as preparation for the "agentic era." ZoomInfo cut 600 people and lost 29% of its stock price as its core product, a contact database, is being repriced by AI systems. These aren't struggling companies cutting costs. They're companies beating expectations and restructuring simultaneously.

The conversation about AI tends to stay at the surface. What can the model do? What gets easier?

The real questions sit below the interface:

• Who controls the infrastructure being built to support this?
• Who absorbs the environmental costs?
• Who loses leverage when systems become essential?
• Who benefits from the acceleration, and who gets displaced by it?

Technology has always reshaped labor and society. What feels different now is the speed at which AI systems are becoming load-bearing before meaningful public conversation has happened about how they should operate.

💭 Consider

Don't spend time beating on a wall, hoping to transform it into a door.

— Coco Chanel

🌱 Final Thought

For years, the pattern was clear. A system scales, something breaks, and institutions catch up slowly. Uncomfortable, but understandable. This week continues to illustrate something different.

The crises aren't arriving sequentially anymore. They're running concurrently. The gap has become wide enough that multiple failures are now the normal operating condition.

That's what institutional fragility looks like when it isn't a crisis. Not a system going dark. A slow erosion of the ground on which decisions are made.

A society notices the fragility of its infrastructure only after it breaks. The harder task is building the capacity to see fragility before the break. That's not a technology skill. It's a judgment skill. And it can't be offloaded to the system you're trying to evaluate.

If these reflections help you think more clearly about education, technology, and the world students are inheriting, there are a few ways to support the work:

• Upgrade to a paid subscription: This gives you access to The Understory and helps keep the main newsletter sustainable.
• Stay as a free subscriber: I'm genuinely glad you're here. The core of Digitally Literate will remain open.
• The Tip Jar: My Ko-fi remains open as I slowly consolidate everything into one ecosystem.

See you next Wednesday on the other side. As always, my email is hello@wiobyrne.com.

🔗 Navigation

Previous: DL 431 • Next: DL 433 • Archive: 📧 Newsletter

🌱 Connected Concepts:

• Institutional Fragility — When digital systems become load-bearing infrastructure for education, their failure is no longer an IT problem — it becomes a pedagogical and institutional crisis.
• The Governance Gap — AI deployment moving faster than democratic oversight means accountability arrives via lawsuits after harm has occurred, rather than through proactive policy.
• SaaS Dependency in Education — The Canvas breach as the predictable outcome of a decade-long reliance on software institutions rent rather than own or control.
• Infrastructure as Curriculum — What students and educators learn about power, dependency, and fragility when the systems they rely on fail publicly.
• AI and Physical Infrastructure — The buildout of AI is reshaping energy grids, water supplies, and labor markets in ways product announcements rarely surface.
• Human Judgment in the Age of AI — Seeing fragility before the break is a judgment skill. It cannot be delegated to the system you are trying to evaluate.

Part of the 📧 Newsletter archive documenting digital literacy and technology.