What the Murakami Persian Translation Race Reveals About AI, Copyright, and Publishing

The Setup: A Copyright-Free Zone Meets AI Speed

Iran is not a signatory to the Berne Convention, which means publishers there are not legally required to secure translation rights before releasing foreign works. This has created a long-standing, openly competitive translation market where multiple publishers race to be first with popular titles.

That dynamic already existed before AI. What’s new is that AI tools have dramatically compressed the timeline.

Murakami’s novel was published in Japan on July 3. Within weeks, three Persian translations had reached the Iranian market. At least three more were in preparation. Some translators completed their versions in under two weeks.

That kind of speed was not possible at scale before AI-assisted translation workflows entered the picture.

The Unofficial Source Text Problem

Here’s where it gets more complicated — and more relevant to anyone thinking about AI in publishing workflows.

Some translators did not work from the Japanese original. Instead, they used an unofficial English version of the novel whose origin is unclear. One translator, Araz Barseghian, noted that this English text showed signs of having been AI-generated.

That creates a layered problem:

• Layer 1: The source text is unofficial and potentially unauthorized.
• Layer 2: That source text may itself be an AI-generated translation of unknown quality.
• Layer 3: Persian translations are then produced from that intermediary, using AI tools for comparison and comprehension.

The result is a published literary work potentially several degrees removed from the author’s original Japanese text, with no clear chain of accountability at any step.

What Translators Are Actually Saying

Not everyone in this race is hiding their methods. Barseghian was notably transparent about his process, describing how he used AI tools to compare the unofficial English text against the official Japanese Kindle edition — checking meaning, pronouns, point of view, and paragraphing.

His framing is worth paying attention to: “The main issue for me is not whether artificial intelligence is used, but transparency and the translator’s responsibility.”

That’s a reasonable position, and it reflects a broader debate happening across the publishing industry right now. AI as a comprehension and comparison tool is different from AI as the final translator. The line between those two uses is blurry, and most publishers don’t yet have clear policies around it.

Ali Mojtahedzadeh, another translator in the race, was more candid about the source issue — acknowledging he obtained an unofficial English translation from a website specializing in such material and completed his Persian version in under two weeks. That admission drew sharp criticism on Persian-language social media.

The Structural Issue AI Is Accelerating

Critics on social media made a point that cuts to the core of this trend: AI didn’t create the underlying problem. The absence of rights enforcement did.

As one observer noted, if rights were properly licensed to a single publisher, there would be no six-way race. AI has simply made the race faster, cheaper, and more accessible to more participants.

This is the pattern worth watching across publishing markets globally:

• Lower cost of entry for translation means more competitors in unregulated markets.
• Faster turnaround means the window for authorized publishers to establish market position shrinks.
• Unofficial source texts become more tempting when speed is the competitive advantage.
• Quality accountability becomes harder to enforce when the production chain is opaque.

What This Means for the Broader Publishing Industry

Iran’s market is an extreme case because of its copyright status, but the pressures it illustrates are not unique to Iran.

Publishers, literary agents, and authors in markets that do enforce copyright are already dealing with questions about AI-assisted translation, the use of machine translation in early drafts, and how to evaluate translation quality when AI is part of the workflow.

The Murakami case adds a specific risk that hasn’t been widely discussed: the AI-generated intermediary text problem. If an unofficial AI translation of a work circulates before an authorized translation exists, it can become the de facto source text for downstream translations in other languages — compounding errors, stylistic choices, and potential hallucinations across multiple editions.

For publishers working in languages where authorized translations are slow to arrive, this is a real and growing exposure tied to intellectual property.

Three Practical Takeaways

1. Source text provenance is now a due diligence issue.
Any publisher or translator using AI tools needs to verify not just the quality of the output, but the legitimacy and origin of the input. An AI-generated intermediary text is not a reliable source, regardless of how fluent it reads.

2. Speed is not the same as quality or legitimacy.
The competitive pressure to publish first is real, but translators and publishers who are transparent about their methods — and who work from authorized source texts — are in a stronger long-term position as scrutiny increases.

3. The “AI as tool vs. AI as translator” distinction matters.
Using AI to check comprehension, flag ambiguities, or compare versions is a different workflow than using AI to generate the translation itself. Publishers and translators who can articulate that distinction clearly will be better positioned as industry norms develop.

The Murakami Persian translation race is not just a publishing curiosity. It’s an early, visible stress test of what happens when AI translation tools meet unregulated markets and unclear source texts. The publishing industry — in every market — should be paying close attention.