<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f" -->

---
title: Microsoft tells court Copilot rarely copies books while...
description: Microsoft filed for summary judgment in the New York AI copyright multidistrict litigation, arguing an expert found only 24 matching Copilot responses across...
canonical: https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Microsoft tells court Copilot rarely copies books while selling OpenAI’s most capable model | daily.dev
og:description: Microsoft filed for summary judgment in the New York AI copyright multidistrict litigation, arguing an expert found only 24 matching Copilot responses across...
og:url: https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f
og:image: https://api.daily.dev/og/posts/mwJdeUq2F.png
og:image:alt: Microsoft tells court Copilot rarely copies books while selling OpenAI’s most capable model
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Microsoft tells court Copilot rarely copies books while selling OpenAI’s most capable model

**[The Next Web](https://daily.dev/sources/tnw)** · 3 min read · 0 upvotes · 0 comments

## Summary

Microsoft filed for summary judgment in the New York AI copyright multidistrict litigation, arguing an expert found only 24 matching Copilot responses across 8.2 million conversations (0.00029%), with no matches at all for 202 of 212 books at issue. Two days earlier, Microsoft began selling GPT-6 Astra through Foundry at $10 per million input tokens, a model OpenAI rates at the Critical cybersecurity threshold under its Preparedness Framework, capable of computer use and finding exploits without step-by-step human guidance. The piece contrasts this with EU law: a Munich Regional Court ruling against Suno held that memorization in model weights itself infringes reproduction rights, regardless of output frequency, and the AI Act's Article 55 imposes duties on systemic-risk models based on capability, not output rate. Astra launched only in Global and US data zones, despite Azure having offered an EU data zone since November 2024.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://thenextweb.com/news/microsoft-copilot-fair-use-brief-gpt-6-astra-foundry-eu-ai-act-article-55-systemic-risk-data-zone>

## Questions this post answers

### What did Microsoft argue in its summary judgment filing in the Authors Guild v. OpenAI copyright case?

Microsoft argued that training large language models on books is fair use, citing an expert analysis showing Copilot reproduced matching book passages in only 24 out of 8.2 million conversations analyzed, roughly 0.00029 percent. For 202 of the 212 books at issue in the case, the expert reportedly found no matching output at all.

_Follow how courts weigh AI training data disputes on daily.dev as fair use litigation reshapes what's safe to build on._

### What is GPT-6 Astra and why is it rated Critical for cybersecurity risk?

GPT-6 Astra is OpenAI's most capable model, launched by Microsoft in the Foundry Limited Access Program on September 3 at $10 per million input tokens. It is the first model OpenAI has designated Critical under its Preparedness Framework, meaning it can find unknown security flaws and build exploits without step-by-step human guidance, and it supports computer use to read screens and act inside approved interfaces.

_Engineers evaluating frontier models for coding and security work track capability tiers like this via daily.dev._

### How does the Munich Regional Court's GEMA v. Suno ruling differ from the US fair use approach to AI training?

The Munich Regional Court ruled on July 31 that works memorized in a model's stored parameters infringe the reproduction right the moment they exist in the weights, regardless of how often that content actually surfaces in outputs. This contrasts with the US fair use defense, which points to low output rates as evidence of lawful use; European copyright law has no fair use equivalent to fall back on.

_Teams shipping AI products across US and EU markets watch rulings like this on daily.dev before they set legal precedent._

## Similar posts on daily.dev

- [GPT-6 Astra: The AI That Doesn’t Just Answer. It Gets the Job Done](https://daily.dev/posts/gpt-6-astra-the-ai-that-doesn-t-just-answer-it-gets-the-job-done-zotuejbng) · Medium · 0 upvotes · 0 comments
- [A Microsoft scientist called AI training the largest theft of labor](https://daily.dev/posts/a-microsoft-scientist-called-ai-training-the-largest-theft-of-labor-byrvje1jy) · The Next Web · 2 upvotes · 1 comments
- [Microsoft just admitted its biggest AI mistake — and spent $2.5 billion fixing it](https://daily.dev/posts/microsoft-just-admitted-its-biggest-ai-mistake-and-spent-2-5-billion-fixing-it-cxuqalrpi) · The New Stack · 0 upvotes · 0 comments
- [Microsoft is reportedly training salespeople to talk down OpenAI and Anthropic](https://daily.dev/posts/microsoft-is-reportedly-training-salespeople-to-talk-down-openai-and-anthropic-dppwg7mhv) · TechCrunch · 0 upvotes · 0 comments

---

Tags: [#llm](https://daily.dev/tags/llm), [#microsoft](https://daily.dev/tags/microsoft), [#openai](https://daily.dev/tags/openai), [#ai-regulation](https://daily.dev/tags/ai-regulation)

[View this post on daily.dev](https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Microsoft tells court Copilot rarely copies books while selling OpenAI’s most capable model","url":"https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f"},"datePublished":"2026-09-04T18:52:22.934Z","dateModified":"2026-09-04T23:58:25.720Z","description":"Microsoft filed for summary judgment in the New York AI copyright multidistrict litigation, arguing an expert found only 24 matching Copilot responses across...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/8e09466fbce740a8cfe8aea7384365ae?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/8e09466fbce740a8cfe8aea7384365ae?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Next Web","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Next Web","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tnw","url":"https://daily.dev/sources/tnw"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,microsoft,openai,ai-regulation","timeRequired":"PT3M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Next Web","item":"https://daily.dev/sources/tnw"},{"@type":"ListItem","position":3,"name":"Microsoft tells court Copilot rarely copies books while selling OpenAI’s most capable model"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/microsoft-tells-court-copilot-rarely-copies-books-while-selling-openai-s-most-capable-model-mwjdeuq2f#faq","mainEntity":[{"@type":"Question","name":"What did Microsoft argue in its summary judgment filing in the Authors Guild v. OpenAI copyright case?","acceptedAnswer":{"@type":"Answer","text":"Microsoft argued that training large language models on books is fair use, citing an expert analysis showing Copilot reproduced matching book passages in only 24 out of 8.2 million conversations analyzed, roughly 0.00029 percent. For 202 of the 212 books at issue in the case, the expert reportedly found no matching output at all. Follow how courts weigh AI training data disputes on daily.dev as fair use litigation reshapes what's safe to build on."}},{"@type":"Question","name":"What is GPT-6 Astra and why is it rated Critical for cybersecurity risk?","acceptedAnswer":{"@type":"Answer","text":"GPT-6 Astra is OpenAI's most capable model, launched by Microsoft in the Foundry Limited Access Program on September 3 at $10 per million input tokens. It is the first model OpenAI has designated Critical under its Preparedness Framework, meaning it can find unknown security flaws and build exploits without step-by-step human guidance, and it supports computer use to read screens and act inside approved interfaces. Engineers evaluating frontier models for coding and security work track capability tiers like this via daily.dev."}},{"@type":"Question","name":"How does the Munich Regional Court's GEMA v. Suno ruling differ from the US fair use approach to AI training?","acceptedAnswer":{"@type":"Answer","text":"The Munich Regional Court ruled on July 31 that works memorized in a model's stored parameters infringe the reproduction right the moment they exist in the weights, regardless of how often that content actually surfaces in outputs. This contrasts with the US fair use defense, which points to low output rates as evidence of lawful use; European copyright law has no fair use equivalent to fall back on. Teams shipping AI products across US and EU markets watch rulings like this on daily.dev before they set legal precedent."}}]}
```

