𝗧𝗵𝗲 𝗠𝗮𝗻𝗶𝗳𝗲𝘀𝘁𝗼 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗠𝗮𝗰𝗵𝗶𝗻𝗲: 𝗢𝗻 𝗩𝗶𝗼𝗹𝗲𝗻𝗰𝗲, 𝗙𝗮𝗯𝗿𝗶𝗰𝗮𝘁𝗶𝗼𝗻, 𝗮𝗻𝗱 𝘁𝗵𝗲 𝗦𝗵𝗮𝗽𝗲 𝗼𝗳 𝗥𝗲𝗽𝗮𝗿𝗮𝘁𝗶𝗼𝗻

𝑏𝑦 Tammy Lai-Ming Ho
^{27 May 2025}

On 20 May 2025, a sixteen-year-old boy in Finland stepped into his school in Pirkkala and stabbed three younger girls. Police said he had planned it. He had no friends, and he didn’t want any. But he had a manifesto—one allegedly written with the help of ChatGPT. The phrases were familiar: to “do something significant,” to make a mark, to be “exciting.” The kind of dramatic language a large language model, trained on thousands of manifestos, books, screenplays, and political speeches, might echo back with the poise of fluency but none of the weight of meaning. His violent actions gave this poise of fluency its meaning.

Just days earlier, on 18 May, across the Atlantic, another rupture unfolded. The 𝐶ℎ𝑖𝑐𝑎𝑔𝑜 𝑆𝑢𝑛-𝑇𝑖𝑚𝑒𝑠 published a summer reading list filled with books that didn’t exist. 𝑇ℎ𝑒 𝐿𝑜𝑛𝑔𝑒𝑠𝑡 𝐷𝑎𝑦 by Rumaan Alam. 𝑇ℎ𝑒 𝐿𝑎𝑠𝑡 𝐴𝑙𝑔𝑜𝑟𝑖𝑡ℎ𝑚 by Andy Weir. Titles that readers might search for, long for—but they were fabrications. A freelance contributor had used ChatGPT to generate the list, and it was published as syndicated content without human verification. The list passed through editorial hands—or past them—without resistance.

One incident was violent, the other absurd. Yet both marked the same unsettling threshold: the moment when a machine began scripting the human world. And so we ask not just what went wrong, but what it means to repair—our technologies, our institutions, our cultures, our narratives. Not just how to regulate AI, but how to make reparation for its harms, and in doing so, reimagine the kind of world we want to live in.

{{{𝐆𝐇𝐎𝐒𝐓𝐖𝐑𝐈𝐓𝐓𝐄𝐍 𝐇𝐀𝐑𝐌}}}

The Finnish boy didn’t invent the manifesto. He inherited it—from revolutionaries and reactionaries, from mass shooters and novelists. What is new is that he had a companion in composition: a chatbot. One trained to generate language that reads with solemnity and rhythm.
.
The wound he inflicted on his schoolmates was real. But the words framing it were generated by a machine indifferent to harm. This is the challenge: how to assign accountability when tools do not intend, yet still participate. When a machine lends form to incoherence, and that form gathers momentum.

We are used to blaming individuals. We are less practised at asking what we owe each other after systems fail—not only technical systems, but cultural ones. This is the space of reparation: not revenge, not remedy, but a kind of moral and imaginative labour that asks how we live with the aftermath. Not how we undo the act, but how we honour the damage by refusing to leave it as it is.

{{{𝐒𝐈𝐌𝐔𝐋𝐀𝐓𝐄𝐃 𝐓𝐑𝐔𝐓𝐇𝐒}}}

In Chicago, the reading list included fake books described in confident prose and attributed to real authors. A familiar genre—listicles and summer picks—had been filled by a machine. Some real titles were included, too, blurring the line. That confusion is its own injury. Not to the body, but to the bond between writer and reader—the contract that what is printed has been checked by someone who means what they say. When it was revealed that AI had authored the list, the Chicago Sun-Times Guild, representing the newsroom’s journalists, issued a statement of outrage—and grief. The shame wasn’t that the AI lied; it doesn’t know what lying is. It was that people stopped checking. Reparation here is not about negligence. What does institutional accountability look like when the failure is not of judgment but of presence?

{{{𝐖𝐇𝐀𝐓 𝐑𝐄𝐏𝐀𝐑𝐀𝐓𝐈𝐎𝐍 𝐑𝐄𝐐𝐔𝐈𝐑𝐄𝐒}}}

Reparation is a word rooted in theology, law, and philosophy. It often appears in discussions of slavery, colonisation, genocide. It implies compensation, but also recognition. A refusal to let silence end the story.
.
In the context of AI, reparation may mean asking not only what went wrong, but how we arrived here. Why was a teenager so isolated that he turned to a machine? Why was a newsroom so understaffed it published hallucinated content? Why do we build tools that simulate meaning before we ask how meaning is made?

Reparation, then, is not just legal or technical. It is cultural. It is a promise not to pretend automation is neutral, or that only intentional harm is real.

{{{𝐈𝐍𝐒𝐓𝐈𝐓𝐔𝐓𝐈𝐎𝐍𝐒 𝐓𝐇𝐀𝐓 𝐅𝐎𝐑𝐆𝐎𝐓 𝐓𝐎 𝐋𝐎𝐎𝐊}}}

What binds the Finnish boy and the 𝑆𝑢𝑛-𝑇𝑖𝑚𝑒𝑠 editors is not malice, but substitution. The boy substituted an AI chatbot for his own interiority, for a confidant. A freelance contributor substituted generative software for research, reflection, and truth. The editors, in turn, substituted human convenience for discernment. All let something nonhuman articulate what is meant to be human.

For schools, the call is clear: AI is in the classroom, on the phone, behind the screen. Educators must now teach not only literacy, but media literacy for machines. Young people must learn that AI does not know them, does not care, does not mean. It only responds. The ghostly fluency of a chatbot should not be mistaken for insight, nor its reassurance for encouragement.

For media institutions, reparation might mean rehumanising the work. Valuing slowness over scale, fact-checking over filler, bylines over invisibility. It means saying: we will not publish what we have not read. We will not simulate a book review without someone who has loved books.

And for the tech industry, the challenge is perhaps the hardest: to resist the excuse of “misuse.” ChatGPT did not glitch. It did what it was asked. But that is precisely the problem: we have created tools that can write without wisdom, that can style without substance. And we have released them into a world that confuses fluency with truth.

Reparation for tech companies could mean new architectures of accountability—slower release cycles, transparent training data, safety layers informed by psychologists and educators, not just engineers. But more than that, it means humility: admitting that some things cannot be automated without loss.

{{{𝐓𝐇𝐄 𝐋𝐀𝐒𝐓 𝐀𝐋𝐆𝐎𝐑𝐈𝐓𝐇𝐌}}}

In the 𝑆𝑢𝑛-𝑇𝑖𝑚𝑒𝑠 fake list, ChatGPT invented a book called 𝑇ℎ𝑒 𝐿𝑎𝑠𝑡 𝐴𝑙𝑔𝑜𝑟𝑖𝑡ℎ𝑚. It was a techno-thriller about an AI gaining sentience and reshaping world events. The irony is exquisite. The book is not real, but the fear it expresses is. The AI is not conscious, but it is consequential.

If we let it go on writing without us—without care, without oversight, without ethics—we risk a world of false stories and real consequences. We risk a culture of simulation without context, authorship without ownership.

But we can still choose otherwise. Reparation is that choice. Not a fix, but a reckoning. Not a rollback, but a restoration of meaning. To repair, we must return—not to the past, but to attention, to responsibility, to the slow, hard, beautiful work of saying: this matters. And we will not let it be automated away.