The Quiet Burning of Our Digital Library
The story of the Library of Alexandria is often told as a catastrophe: flames, conquest, a single dramatic moment in which centuries of human knowledge vanished. But historians are far less certain about such a clean ending. What they largely agree on is something more troubling. The greatest library of the ancient world did not disappear overnight. It faded. Through neglect. Through political pressure. Through shifting priorities. Through a slow erosion of support, until one day people realized that the texts they needed were simply no longer there.
We tend to remember Alexandria as a tragedy of destruction. It was, more accurately,a tragedy of unnoticed loss.
Something uncomfortably similar is beginning to happen to our digital memory.
From Knowledge to Infrastructure
Today’s equivalent of Alexandria is not a single building or institution, but a set of infrastructures that quietly preserve the public record of the internet. Chief among them is the Internet Archive’s Wayback Machine: a tool that allows journalists, researchers, lawyers, and citizens to see and safeguard what was published online at a specific moment in time.
This matters because the web is not stable. Pages are edited. Statements are softened. Policies disappear. Tweets are deleted. Entire sections of government or corporate websites quietly vanish. In this environment, memory is not a given; it has to be actively preserved.
Over time, web archives have become essential investigative tools. They allow journalists to follow what was deleted, what was altered, what was quietly rewritten, and what was removed without explanation. They help reconstruct timelines, expose contradictions, verify claims, and hold powerful actors to account. Courts increasingly rely on archived webpages as evidence, recognizing that digital memory is now part of the legal and democratic infrastructure.
This function transcends mere technological preservation or bittersweet nostalgia: it constitutes the infrastructure of democratic accountability itself.
Fear, AI, and the Shrinking of Memory
And yet, this infrastructure is now under strain. Not because its purpose is disputed, but because it has become entangled in a very different fear: artificial intelligence.
As AI companies scrape the web to train large language models, publishers are understandably alarmed. Their business models, already fragile after decades of disruption, now face a new threat: the wholesale ingestion of their archives into training datasets. In response, many have tightened access controls, blocked crawlers, and restricted the Internet Archive’s ability to capture and display their pages.
Major publishers, including The Guardian, The New York Times, the Financial Times, and USA Today, have hinted they’re examining the Internet Archive’s access to their content. From the publisher’s perspective, this is understandable. If AI companies are looking for large, structured datasets, archives begin to look less like libraries and more like liabilities. News publishers have referred to the Internet Archive as an involuntary “back door” to their catalogues, allowing tech companies to continue scraping their content.
But this is where the Alexandria parallel becomes uncomfortable.
The erosion of the ancient library did not happen because scholars stopped valuing knowledge. It happened because the institutions that sustained it were no longer aligned with political and economic incentives. Preservation became someone else’s problem, until it was no one’s problem.
Blocking archives to protect against AI does not just limit machine access. It limits human memory.
When Evidence Becomes Optional
The consequences of this shift are not abstract.
Disinformation thrives in environments where verification is difficult. It does not require better lies, only weaker evidence. When archived webpages disappear or become inaccessible, it becomes harder to prove that a statement was made, that a promise existed, or that a narrative changed.
Investigations into climate data being scrubbed from government websites, public history being rewritten, or policies quietly altered all rely on the ability to point to what existed before. Without archives, these debates become exercises in assertion rather than evidence.
The tragedy of Alexandria is not only that knowledge was lost, but that we cannot fully know what was lost. We cannot consult the missing texts. We cannot test interpretations against originals. We are left arguing in the dark.
A web without reliable archives recreates that condition, at digital speed.
When the Truth Is Harder to Reach Than the Lie
This problem is compounded by a structural asymmetry in today’s information environment.
High-quality journalism, academic research, court records, and historical material are increasingly locked behind paywalls, subscriptions, or institutional access. Meanwhile, misinformation, conspiracy theories, and ideologically driven content circulate freely, optimized for reach rather than accuracy.
As Current Affairs editor Nathan J. Robinson famously observed: “the New York Times, the New Yorker, the Washington Post, the New Republic, New York, Harper’s, the New York Review of Books, the Financial Times, and the London Times all have paywalls. Breitbart, Fox News, the Daily Wire, the Federalist, the Washington Examiner, InfoWars: free!”
Web archives partially correct this imbalance. They lower the cost of verification. They allow people without subscriptions, credentials, or institutional backing to check what was actually said. They make truth easier to reach.
Restricting those archives in response to AI fears reverses that dynamic. It raises the cost of fact-checking at precisely the moment when speed and accessibility determine which narratives dominate.
In Alexandria, knowledge became scarce not because it ceased to exist, but because access to it narrowed. The same risk now applies online today.
AI Didn’t Break Copyright, It Exposed a Fault Line
None of this is because copyright law suddenly stopped working. AI did not break copyright. It exposed how dependent the system has always been on assumptions of human scale: slow creation, limited distribution, manageable enforcement.
At AI scale, those assumptions collapse. Faced with that collapse, we are responding by extending licensing and control mechanisms not only to commercial exploitation, but to preservation itself.
This is a familiar story. Libraries have long been pushed from ownership to licensing, from stewardship to permission. Applying that logic to web archives turns memory into a negotiable commodity. And memory does not survive long under licensing terms.
The Library of Alexandria was not destroyed because its contents lacked value. It was lost because preserving them no longer aligned with power, politics, or profit.
What Is Actually Worth Defending
If the goal is to fight disinformation and preserve democratic accountability, then weakening preservation infrastructure is a profound mistake.
What needs defending is not uncontrolled data extraction, but the distinction between public-interest archiving and commercial exploitation. Between libraries and profit machines. Between memory and mining.
That means:
- Targeting abusive, large-scale AI scraping directly;
- Protecting library and archival exceptions that allow preservation for public interest;
- Recognizing web preservation as civic infrastructure, not a data leak; and
- Resisting fear-driven responses that sacrifice long-term memory for short-term control.
Governance that undermines our ability to verify the past does not make society safer. It makes it easier to manipulate.
The Quiet Lesson of Alexandria
The loss of the Library of Alexandria is remembered not because of malice, but because of consequence. No one set out to erase centuries of knowledge. Yet history was shaped by what could no longer be consulted, challenged, or remembered.
Today, we risk repeating that mistake in a quieter form.
Not through fire, but through access controls.
Not through conquest, but through fear.
Not through a single act, but through many reasonable decisions whose cumulative effect is erasure.
By the time we realize what is missing, it may already be too late to recover it. And like Alexandria, we will be left arguing not about what happened, but about what we can no longer prove ever existed.
Get the article’s core concepts in a unique auditory summary—listen to the AI-generated track below!
This song was created entirely using artificial intelligence tools. If you enjoy this experiment, keep an eye out for each song illustrating our December articles.
Written byCaroline De Cock, LL.M., Head of Research
