The New York Times has filed a motion to amend its copyright infringement lawsuit against OpenAI and Microsoft, shifting focus more heavily on Microsoft after a Supreme Court ruling raised the bar for contributory infringement claims. The amended complaint was filed in a federal court in New York on June 25, 2026, three months after the Supreme Court's decision changed the standards for these types of cases.

The original lawsuit was filed on December 27, 2023, making it one of the longest-running AI copyright cases in US legal history. The updated filing comes after the Supreme Court sided with Cox Communications in a separate case where Sony attempted to hold Cox liable for music piracy on its network. That ruling established a stricter standard for contributory infringement, requiring plaintiffs to prove that a party intentionally acted to induce illegal conduct rather than simply knowing infringement was occurring.

In response to this precedent, the Times voluntarily dropped claims that OpenAI helped its users break copyright rules, recognizing that the legal standard has changed and plaintiffs must now prove intentional encouragement of illegal conduct. The newspaper also voluntarily dismissed two other claims against all defendants as part of the filing.

The amended complaint alleges that Microsoft actively encouraged copyright infringement by building a custom supercomputing system specifically designed to enable large-scale training on copyrighted material. The Times argues that this infrastructure, which reportedly includes over 285,000 CPU cores and 10,000 GPUs, was not generic cloud computing that OpenAI happened to rent. Instead, the newspaper alleges it was a tailored system built specifically to enable model training that consumed millions of Times articles without permission or payment. The complaint asserts that Times articles were disproportionately weighted in the training data to help the models mimic high-quality reporting.

The Times alleges that ChatGPT produces near-verbatim excerpts of its copyrighted works, sometimes when users ask to see the next paragraph to get around paywalls and sometimes without any special prompting. The complaint also cites cases where models falsely attribute fabricated content to the Times, including fake quotes and a fabricated article linking orange juice to lymphoma that was never published. The newspaper argues these outputs directly substitute for its own product, causing market harm and reputational damage, and that Microsoft has profited by integrating the Times-trained models across its product line, helping boost its market capitalization by a trillion dollars over the past year.

Microsoft dismissed the amended complaint as a "last-ditch effort" to rescue a claim weakened by recent legal precedent. OpenAI maintains that its models are trained on publicly available data and grounded in fair use. The Times continues to seek a permanent injunction against further infringement and extensive damages, claiming the defendants have wrongfully profited from copyrighted works they do not own.

The next key moment in the case will be whether the court grants the Times' motion to amend. If approved, Microsoft would face a significantly more targeted legal challenge than the one it has been defending against since 2023. The case has significant implications for the AI industry and investors, as Microsoft has positioned its Azure cloud platform and its deep OpenAI partnership as central components of its artificial intelligence growth strategy.

Original Sources/Tags: arstechnica.com, nytimes.com, news.bloomberglaw.com, arstechnica.com, cryptobriefing.com, news.bloomberglaw.com, bostonherald.com, pymnts.com, (openai), (microsoft), (sony), (chatgpt), (copyright), (journalism)

This article provides limited actionable information for a normal person. It reports on a lawsuit between the New York Times and technology companies over artificial intelligence training, but it does not give clear steps, choices, or tools a reader can use right now. There are no specific resources mentioned that an individual can access or act upon. A person reading this cannot influence the lawsuit, verify the legal claims, or use the information to make a concrete decision about their own use of artificial intelligence tools. The article gives the reader very little to do.

The educational depth is modest. The article mentions that artificial intelligence models can reproduce copyrighted articles, that courts consider fair use and contributory infringement, and that companies build powerful computers to train these models. However, it does not explain how courts actually decide fair use in AI training cases, what evidence standards apply when one side claims intentional copying, how training data weighting works technically, or what specific changes would make AI training safer for content creators. The information stays at the surface level of reporting legal claims without teaching the reader how to understand the technology or the legal tradeoffs involved.

Personal relevance is moderate for people who create written content online or who rely on journalism for their work. For someone who does not publish articles or whose livelihood does not depend on copyrighted written work, the information has limited effect on their daily safety, money, health, or responsibilities. The article does not explain how to protect your own written work from being used in AI training, how to check whether a product was trained on your content, or what questions to ask before using an AI tool that summarizes news.

The public service function is weak. The article does not warn any specific population about a danger in a way that helps them act. It notes legal claims about market harm and copyright infringement but does not provide guidance on how creators can protect their work, how consumers can choose AI tools that respect copyright, or how to evaluate whether a product respects the rights of journalists and writers. It exists mainly as a summary of a legal filing rather than as a service to help people act responsibly.

There is no practical advice in this article for an ordinary reader to follow.

The long term impact of reading this is small for personal action. It may slightly increase awareness that AI training raises copyright concerns and that companies sometimes build powerful systems specifically to process large amounts of written content. It does not give the reader tools to evaluate similar claims critically in future news cycles or to apply lasting principles when judging new AI products.

The emotional impact leans toward mild concern without offering any constructive response. The article describes legal claims about theft and market harm but does not balance these with practical guidance about how to handle similar situations, how to evaluate whether an AI tool respects your own work, or how to make informed decisions about using products trained on copyrighted material. The reader is left aware of a problem without gaining tools to respond to it.

The language is measured and not overtly clickbait. Phrases like "bespoke supercomputing system" and "last-ditch effort" add some weight but do not sensationalize. The article does not use exaggerated numbers or false claims. It does frame the Times's position through detailed allegations while giving only brief space to the defendants' responses, but this is a limitation of sourcing rather than a deliberate attempt to shock.

The article misses several chances to teach broader lessons. It could explain how readers can evaluate whether an AI tool respects copyright, what questions to ask about training data and data sources, how to check whether a product reproduces copyrighted material, or what general principles apply to any service that processes large amounts of written content. It could also explain how companies respond to legal pressure and how to recognize when a lawsuit reflects a genuine legal dispute versus a strategic move.

A person who wants to keep learning can use basic reasoning methods without relying on external data sources. Compare claims by checking whether multiple independent legal experts agree on how the law applies to AI training. Examine patterns by watching whether other publishers have filed similar lawsuits and what outcomes resulted. Consider general principles. When a company says training on public data is legal, ask whether the data was truly public or behind a paywall, and whether the use competes with the original source. These questions require only common sense.

Here is concrete guidance based on universal principles that readers can apply regardless of location. When you are considering using an AI tool that summarizes or generates text, find out what data it was trained on, whether that data included copyrighted material, and whether the tool gives credit to original sources before you rely on it. If you create written content online, take simple steps to understand your rights by reading the terms of service of platforms where you publish, keeping records of your original work, and learning how copyright law in your country protects written content. When you hear about a lawsuit between a publisher and a technology company, compare what both sides say rather than accepting one side's framing, and look for independent legal analysis rather than relying on press releases. When a company tells you that its use of data is legal, ask yourself whether the original creators consented, whether they are compensated, and whether the use competes with the original product. If you want to support journalism and original reporting, develop simple habits like subscribing to news sources you value, sharing links to original articles rather than copying text, and choosing AI tools that credit or compensate content creators. Clear, documented, supported efforts to understand your own rights and choices are more effective than relying on a company's assurances alone.

The text says Microsoft "actively encouraged the theft of its works." The word "theft" makes the act sound like a clear crime, but copying data for AI training is a legal dispute about fair use, not proven theft. This trick helps the Times by making Microsoft look like a criminal instead of a company in a legal disagreement. It hides the real legal question by using a strong word that pushes bad feelings.

The text says Microsoft built the system "specifically to train large language models on copyrighted journalism without permission." The word "specifically" makes it sound like Microsoft's main goal was to steal news, hiding that the computer was built to train AI on all kinds of data. This bias helps the Times by making Microsoft look fully driven by bad intent. It leaves out that news was only one part of what the system was meant to do.

The text says Times articles were "disproportionately weighted in the training data." This word choice makes it seem like the Times was specially targeted and hurt more than others. The bias helps the Times by making its harm seem bigger and more unfair than just being part of a large mix of data. It hides whether other sources were also weighted heavily or if this is just how AI training works.

The text says AI models are "falsely attributing fabricated content to the newspaper." The words make it sound like OpenAI purposely creates lies and sticks them on the Times to hurt it. This trick helps the Times by making AI errors look like an active attack rather than known software mistakes. It hides that AI often mixes up sources by accident without any bad intent.

The text says these outputs "directly substitute for its own product, causing market harm." The phrase pushes readers to believe people no longer need to buy Times articles because ChatGPT gives them away for free. This bias helps the Times by stretching what an AI summary does into total lost sales. It hides that an AI giving parts of an article might not replace reading full news reports.

The text says Microsoft profited from this, "helping boost its market capitalization by a trillion dollars over the past year." Linking one lawsuit's claims to a huge trillion dollar gain tricks readers into thinking this stolen news is worth that much money. The bias helps the Times by making its stolen work seem incredibly valuable and Microsoft's profit seem deeply tied to it. It hides that most of Microsoft's huge money gain comes from many other things besides this one case.

Microsoft dismissed the complaint as a "last-ditch effort" while OpenAI maintains training is "fair use." These short phrases give very little room for their side compared to all the space given to explain how they stole and hurt others. The setup shows bias because it lets one side tell a long story while barely letting defenders speak for themselves at all in their own words

The text conveys a strong sense of outrage and moral indignation on behalf of the New York Times, which is evident through the choice of words like "theft" and "actively encouraged." These words are not neutral; they suggest deliberate wrongdoing and criminal behavior, framing Microsoft's actions as intentional harm rather than a legal dispute. This emotion is intense and serves to paint Microsoft as a villain that knowingly stole valuable work, pushing the reader to feel sympathy for the Times and anger toward the tech companies. The purpose is to create a clear victim-and-bully dynamic, making the newspaper seem like a defender of justice against a powerful, greedy corporation.

A feeling of alarm and worry is also present, particularly in the description of the AI system as "specially built" and "one of the most powerful in the world" for the purpose of training on copyrighted material without permission. This language suggests a large-scale, unstoppable threat that could harm not just the Times but all journalism. The mention of AI models "falsely attributing fabricated content" adds fear about reputational damage and the spread of lies, while the claim that outputs "directly substitute for its own product" raises concern about economic harm. These emotions are meant to make the reader anxious about the future of trustworthy news and the power of big tech, potentially building support for the Times's legal action.

The text also expresses a tone of determination and resolve through the Times's actions, such as seeking a "permanent injunction" and "extensive damages." This conveys strength and persistence, showing the newspaper as willing to fight back rather than accept defeat. The emotion here is firm and confident, designed to build trust in the Times as a principled fighter for what is right. It inspires respect and possibly encourages others to see the lawsuit as a necessary stand against injustice.

Microsoft's response introduces a contrasting emotion of dismissal and contempt, calling the complaint a "last-ditch effort." This phrase suggests desperation and weakness, implying the Times's case is failing and the newspaper is grasping at straws. The emotion is belittling and serves to undermine the Times's claims by making them seem like a strategic trick rather than a legitimate grievance. OpenAI's stance, described as calmly maintaining "fair use," conveys confidence and rationality, positioning the tech companies as reasonable and law-abiding compared to what they frame as the Times's emotional overreaction.

The writer uses persuasive tools like extreme language to heighten emotional impact. Words like "theft," "bespoke," "disproportionately," and "trillion dollars" are chosen to sound dramatic and shocking, making the situation seem bigger and more urgent than a simple legal disagreement. The comparison between the Times's detailed allegations and the defendants' brief responses creates an imbalance that favors the newspaper's story, steering the reader to view its claims as more credible and important. The repetition of the idea that Microsoft built a powerful system specifically for this purpose reinforces the sense of intentional harm, while the focus on market harm and fabricated content highlights the potential consequences, making the reader more likely to side with the Times out of concern for journalism and fairness.