Back
In AI Training Dispute, Court Rules Corporations Can Invoke 'Private Use' Exception
In a groundbreaking ruling that redefines the contours of copyright law in the age of artificial intelligence, a High Court has held that the statutory exception for "" extends to corporations, not just individuals. The decision, which arose from a dispute over 's use of copyrighted news articles for training its language models, marks a significant departure from conventional interpretations and has sent ripples through the legal community.
The Background: AI Training and Copyright
At the heart of the case was the question whether , the developer of ChatGPT and other generative AI systems, could lawfully copy and process copyrighted works—specifically, articles from —without a licence. The company relied on the "" exception found in many copyright statutes, which traditionally permits reproduction of works for strictly non-commercial, individual purposes. argued that its internal training process, which involves scanning and storing digital copies in secure servers inaccessible to the public, qualifies as "private use". The appointed by the court countered that the phrase must be read as a compound expression—""—where the words draw meaning from each other under the interpretative principle of . Under that reading, "private" would carry a connotation similar to "personal", i.e., limited to a natural person acting for themselves.
The Judgment's Core Reasoning
The court rejected the amicus's argument and adopted a far broader construction. Relying on dictionary definitions, the court observed that the word "private" means
"restricted to a particular person, group, or class"
. Since
holds the ANI works in a closed, access-controlled environment without any public dissemination, the court deemed this "private use". The judgment explicitly stated that adopting the
rule in the manner suggested by the amicus would render the word "private" otiose—pointless—because if it meant the same as "personal", it would not have been included separately. By treating "private" as a standalone term, the court paved the way for any entity—whether a multinational corporation or a single individual—to claim the exception, provided the use occurs in a non-public space.
Critique of the Reasoning
Legal experts have been quick to point out the potential logical flaws in the court's approach. The conventional understanding distinguishes "personal use" (use by an individual for their own purposes) from "private use" (use in a domestic or non-commercial setting). The classical rule has been that the two terms overlap and together carve out a narrow
for ordinary, non-industrial activities. To stretch "private" to encompass massive, industrial-scale data copying by corporations turns that settled principle on its head. As one commentator noted,
"Personal use simply means using something for yourself, while private use means using it in your own private space. The law makes sense without having to stretch the word 'private' to justify massive data copying by corporations."
The court's response to the argument is also open to question. The phrase "" appears in many statutory or provisions. Courts have historically treated it as a compound exception covering only the truly non-commercial sphere. By disaggregating the two adjectives, the judgment creates a circularity: if any group, including a corporation, can have "private" use simply by keeping the copies behind a firewall, then the exception would swallow the rule. Copyright owners would find their works freely copied and mined for commercial AI training without compensation, as long as the AI company does not publicly distribute the copies.
Legal Implications and Impact on Practice
For intellectual property practitioners, this ruling introduces a new layer of unpredictability in copyright litigation involving AI. Until now, the predominant view was that systematic, large-scale copying of copyrighted content to train commercial AI models constitutes infringement unless covered by specific text‑and‑data‑mining exceptions—which most jurisdictions have introduced precisely to avoid this exact ambiguity. The judgment's expansion of "private use" to corporations effectively creates a judicial that may pre‑empt legislative text‑and‑data‑mining provisions and undermine the carefully crafted balancing of interests.
The decision also raises practical questions for corporate legal departments. If a company can now claim that its internal data storage and processing are "private use", the boundaries of copyright become dangerously porous. For instance, a corporation could digitise entire libraries of copyrighted books or databases for internal analytics and machine learning, arguing that it is not distributing the copies publicly. The ruling thus blurs the line between private copying (exempt) and (actionable). Litigants in future cases will likely test the limits of this reasoning, potentially leading to a flood of applications seeking to quash infringement claims on the basis of "private use" alone.
Conclusion
The High Court's judgment marks a significant shift in copyright jurisprudence, one that favours the interests of AI developers and other technology companies at the expense of content creators. By anchoring the definition of "private" to the physical or virtual space where copies are held, rather than to the nature or commerciality of the user, the court has opened a new frontier of copyright law. Whether this interpretation survives appellate scrutiny remains to be seen, but for now, the AI industry has gained a powerful argument to defend its data‑sourcing practices. The legal community will watch closely as further challenges—and likely legislative reactions—emerge in the wake of this transformative ruling.
