Risks of Copyright Infringement for Businesses

The largest copyright settlement in U.S. history wasn’t about music or movies. It was about AI training data. Anthropic will pay $1.5 billion and destroy datasets sourced from Library Genesis and Pirate Library Mirror. That number isn’t just a penalty. It is a warning label for every enterprise buyer of LLMs. What CIOs, CTOs, and GCs need to know: 1. Dataset provenance is now a board-level risk. The court drew a bright line: purchased-and-scanned books may qualify as fair use, but pirated books are “irredeemably infringing.” If your vendor cannot show data lineage, you are buying litigation risk. 2. Past and future liability are different. Anthropic only secured a release for past training. Any outputs from Claude, and any future ingestion of copyrighted material, remain open to lawsuits. If you deploy an LLM, ask how liability for outputs is handled in contracts. 3. There is no blanket amnesty. The release applies only to a defined list of about 500,000 works. Everything else remains actionable. Enterprises should not assume one settlement protects them from the next. 4. AI costs are about to change. Three thousand dollars per book became the reference price for copyrighted works in training datasets. That is far above statutory minimum damages. Expect higher licensing costs to be built into the price of frontier models. 5. Governance must extend to AI supply chains. Just as enterprises built third-party risk programs for cloud and SaaS, they will now need AI governance frameworks: • Data provenance checks in procurement • Output indemnities in vendor contracts • Ongoing monitoring for infringement claims Bottom line for buyers: The age of “scrape first, ask later” is over. Enterprises that treat AI adoption as a procurement and governance problem, not just a technology problem, will be the ones who stay protected while others are dragged into court.

Anthropic agreed to pay at least $1.5 billion plus interest to settle a copyright lawsuit over its alleged use of millions of pirated books to train its large language models. This is one of the largest IP settlements in history. Here’s what happened: The class-action lawsuit represented authors of up to 7 million books. Anthropic allegedly downloaded copyrighted works from pirate websites without permission. Under the settlement, authors will receive roughly $3,000 per book for 500,000 books in the class, with the payout increasing if more claims are submitted. The company also agreed to destroy the illegally downloaded data. While this is a landmark settlement in the AI space, it is also worth unpacking. Here’s why it matters: 𝐖𝐡𝐚𝐭 𝐀𝐧𝐭𝐡𝐫𝐨𝐩𝐢𝐜 𝐝𝐢𝐝 𝐰𝐫𝐨𝐧𝐠 Using pirated content for AI training, even if not ultimately incorporated into the model, counts as copyright infringement. No proper licenses were obtained, unlike some competitors who proactively license content from publishers. Systematic, large-scale copying triggered the risk of damages, reportedly up to $1 trillion. 𝐈𝐦𝐩𝐥𝐢𝐜𝐚𝐭𝐢𝐨𝐧𝐬 𝐟𝐨𝐫 𝐨𝐭𝐡𝐞𝐫 𝐀𝐈 𝐜𝐨𝐦𝐩𝐚𝐧𝐢𝐞𝐬 IP compliance is now non-negotiable. AI startups can no longer rely on scraping content indiscriminately. Licensing may become the industry standard, as seen with OpenAI and media publishers. Class actions are real risks as even companies with huge valuations are not immune. 𝐊𝐞𝐲 𝐭𝐚𝐤𝐞𝐚𝐰𝐚𝐲𝐬 𝐟𝐨𝐫 𝐜𝐨𝐧𝐭𝐞𝐧𝐭 𝐠𝐞𝐧𝐞𝐫𝐚𝐭𝐢𝐨𝐧 Always verify that your datasets or training materials are licensed, public domain, or ethically sourced. Conduct IP audits to ensure your AI models are not infringing on copyrighted works. Consider the reputational and financial risks: cutting corners can cost more than proper licensing. 𝐂𝐨𝐧𝐭𝐞𝐱𝐭 𝐦𝐚𝐭𝐭𝐞𝐫𝐬 The settlement is a partial victory for Anthropic: the payout is less than 1% of its valuation, and it didn’t have to break its AI models. Still, it’s a strong warning shot to the AI ecosystem: respect copyright, or risk billions. Anthropic’s case shows that AI innovation cannot come at the expense of creators’ rights. For founders, investors, and developers: ethical and legal diligence is now a competitive advantage, not just a compliance checkbox. Lesson for the AI community: When generating content, building datasets, or training models, don’t take shortcuts that could cost you the company. Article: Bloomberg. Opinion: Mine

The AI Copyright Reckoning: What Disney's Lawsuit Means for Your Business The entertainment industry just fired its biggest shot yet in the AI copyright wars. The Walt Disney Company and NBCUniversal have teamed up to sue Midjourney, marking the first time major Hollywood studios have taken legal action against a generative AI company. Combined with Getty Images' copyright lawsuit against Stability AI that began at London's High Court this week, we're witnessing a critical moment that will reshape how AI companies operate. The Disney-NBCUniversal lawsuit claims Midjourney pirated their libraries to generate "endless unauthorized copies" of characters like Darth Vader and the Minions. Getty Images accuses Stability AI of misusing over 12 million Getty photos to train its Stable Diffusion system. Both cases represent a clear escalation from individual artist lawsuits to major corporate copyright holders taking direct action. AI companies argue they're operating within existing legal frameworks and that overly broad copyright interpretations could stifle innovation. They contend their systems learn patterns to create new, original works rather than storing exact copies. However, courts are showing increased willingness to let these cases proceed, with one California judge recently indicating he was inclined to green-light a copyright lawsuit against multiple AI companies. These disputes have the potential to reshape copyright licensing in the AI age and create new legal frameworks that could materially impact AI development costs and market access. We're likely to see increased demand for properly licensed training data, creating new revenue streams for content creators but higher barriers for AI companies. For businesses, the message is clear--the Wild West era of AI training data is ending. Companies using AI tools should audit their current usage and understand the training data sources behind their systems. Those developing AI systems must invest in proper data licensing rather than relying solely on scraped internet data. The cost of licensing may be less than the potential liability from copyright infringement claims. Companies that adapt by building proper licensing relationships and respecting intellectual property rights will be best positioned for long-term success, while those that ignore these developments do so at their own peril. What steps is your organization taking to navigate AI copyright challenges?

Let's talk photos. There are too many business owners out there that still do not understand that they cannot just grab a photo from the internet and use it to promote their business. All photos are creative works fixed in a tangible medium and therefore, they are copyright protected by the photographer that took them. ALL OF THEM. And most photographers actually go the extra step of securing a copyright registration. The internet is not your playground. If you own a business, you must be using others' works responsibly. That means with permission and a license. Last week, I saw a business owner complaining on social media that they received a C&D letter from a copyright owner about a photo they had not properly licensed. The business owner admitted they had not licensed the photo. After receiving the C&D letter, they took the photo down and he didn't understand why they were liable for any damages because they took it down. He believed the C&D letter was a warning and since he complied, there was nothing more to do. WHAT? THIS IS 100% FALSE. First, no one is required to send you a C&D. If they do, that's usually not a good sign. There is no such thing as a good C&D. Second, they used the photo in connection with their business for who knows how long. That copyright owner deserves to be compensated! Third, you don't get to skate thru life "borrowing" others' works until you get caught and then you will fix it. What kind of strategy is that?? If you were pulled over for speeding, would you say to the police officer, "yeah, I get it. Thanks for catching me doing something illegal. Scouts honor, I won't speed the rest of the way home?" Of course not, you'd get a ticket and you'd have to pay a fine! The same applies to all laws - trademark or otherwise. Here's what you do if you're in this situation. You purchase a package of photos from the many subscriptions out there or use royalty free photos. If you choose royalty free photos, be sure to READ THE LICENSE and understand whether or not it covers your use. Pulling a random photo from Google is not a royalty free licensed photo. EVER. Your better bet? Just hire an attorney so you don't have to pay those shakedowns. Are they trolls? YES! But, they have the law on their side, and it's the easiest gig in the world for an attorney to send those letters out. If you get one, speak to an attorney. They are trained to help you. Ignorance of the law is not an excuse. #brandprotection #copyrightlaw #copyrightinfringement

Dr. Saju Skaria's Weekly Reflections: 20/2023 Intellectual Property (IP) issues and Generative AI The use of Generative AI has taken the industry by storm. I thought it’s imperative to touch on this critical issue, i.e., intellectual property, before we close the discussions on AI. AI technology, leveraging data lakes and question snippets to recover patterns and relationships, is helping immensely in creative industries. However, we have a critical issue that the legal fraternity is trying to address: copyright infringement, ownership of AI-generated work, and leveraging unlicensed content in training data. Trained AI tools can replicate copies of original work (for example, paintings or photographs), which is a copyright infringement. A further challenge is that the users might create copies of the original that need to be more transformative, thereby causing the credibility of the original work. These unauthorized derivatives can cause significant penalties, and the courts are already dealing with such issues. There is significant debate around the “fair use doctrine” that allows reviewing the copyrighted without the owner’s permission for purposes like criticism (including satire), comment, news reporting, teaching (including multiple copies for classroom use), scholarship, or research,” and for transformative use of the copyrighted material in a manner for which it was not intended. A word of caution for companies is how to use Generative AI and leverage content. It’s a tightrope walk. Not even accidentally using copyrighted content, directly or unintendedly, without adequate protection can cause significant penalties. How could we reduce the risk of getting stuck in an IP violation? Here are a few recommended steps. 1.     AI developers (individuals /organizations) must ensure that they comply with the law regarding acquiring data to train their models. 2.     Creators, both individual content creators and brands that create content, should take steps to examine risks to their intellectual property portfolios and protect them. 3.     Businesses should evaluate their transaction terms to write protections into contracts. As a starting point, they should demand terms of service from generative AI platforms that confirm proper licensure of the training data that feed their AI. Finally, with appropriate protection, businesses can build portfolios of works and branded materials, meta-tag them, and train their generative AI platforms to produce authorized, proprietary (paid-up or royalty-bearing) goods as sources of instant revenue streams. I welcome your thoughts and views on the topic. #AI #Leadership Bharat Amin, NACD.DC ML Kabir Sandeep (Sandy) M. Krishnan CA Randhir Mazumdar Dr. Swati Karve, PhD Psychology Ashish Saxena Shiny Skaria