Tobias Koprowski’s Post

🚨 BREAKING: 🇪🇸 Spain proposes a draft Royal Decree regulating COLLECTIVE LICENSES for the mass exploitation of COPYRIGHTED works to train AI. 🔥 This could SHAKE UP the field of AI; here's what everyone should know: ➡️ According to the draft decree's summary: "The purpose of the draft Royal Decree is to develop article 163 of the TRLPI (relating to the granting of non-exclusive authorizations for the use of the repertoire of management entities), with the aim of facilitating the granting of said non-exclusive authorizations (or collective licenses) in the context of technological development of AI (and, in particular, for the development of general-purpose AI models). The specific mechanism for this would be extended collective licenses, an instrument provided for in Article 12 of the EU Copyright Directive (2019/790), whose inclusion in the internal regulations of the EU Member States is voluntary but which has proven particularly timely in the context of AI." ➡️ The first article of the proposed decree establishes the conditions for the concession of these expanded collective licenses: ╰┈➤ "Obtaining intellectual property rights holders' authorization on an individual basis is so onerous and difficult that it makes the required operation improbable. ╰┈➤ The collective management entity is, on the basis of its mandates, sufficiently representative in Spain of the rights holders of the corresponding category of protected works or services and of the rights that are the subject of authorization, which will be accredited by the certificate of representation referred to in article 3. ╰┈➤ All rights holders are guaranteed equal treatment in relation to the terms of the non-exclusive authorization. ╰┈➤ Rights holders who have not authorized the entity to grant the non-exclusive authorization may exclude their protected works or services from the extended collective license at any time, easily and effectively. ╰┈➤ Appropriate publicity measures are taken to inform rights holders who have not granted a management mandate, for a reasonable period, before the protected works or services are used under the non-exclusive authorization." ➡️ This is still a draft proposal and might change. However, given that it's based on Article 12 of the EU Copyright Directive, if it works, other EU countries might follow suit. ➡️ This is an extremely interesting development in the field of AI copyright, which might signal how the field might evolve and how lawsuits might be decided in the months to come, at least from an EU perspective. ➡️ If you want to have a say on AI copyright issues in the EU, I highly recommend that you submit your contributions and participate in this discussion. The deadline is December 10. If you have friends who might want to participate, share it with them. 👉 Link below. 👉 Never miss my AI updates: join 40,000+ people who subscribe to my newsletter (below). #AI #AICopyright #AIRegulation #AIGovernance #CollectiveLicenses

💣 The EU AI Act and IP: here are the key points. General-purpose models (GP-AI - think GenAI) are said to present "challenges to artists, authors, and other creators and the way their creative content is created, distributed, used and consumed" - Recital (105). Providers of GP-AI must have "a policy to comply with Union copyright law", including complying with opt-outs from the EU's commercial TDM exception using "state of the art technologies" - Recital (105) and Article 53(1)(c). Radically, it appears Union copyright law must be followed even where a GP-AI is trained outside of the EU: "Any provider ... should comply with this obligation, regardless of the jurisdiction in which the copyright-relevant acts underpinning the training of those general-purpose AI models take place. ... no provider should be able to gain a competitive advantage in the Union market by applying lower copyright standards than those provided in the Union." - Recital (106). Providers of GP-AI must disclose details of the content used for training - Recital (107) and Article 53(1)(d). No exception is made for open source models - Recital (104). Providers fine-tuning GP-AI need to disclose the new training data sources - Recital (109). The scope of the public disclosure is to be decided. Recital (107) says: "While taking into due account the need to protect trade secrets and confidential business information, this summary should be generally comprehensive in its scope instead of technically detailed to facilitate parties with legitimate interests, including copyright holders, to exercise and enforce their rights under Union law, for example by listing the main data collections or sets that went into training the model, such as large private or public databases or data archives, and by providing a narrative explanation about other data sources used." Under Article 56(2)(b), the AI Office needs to set out a code of practice covering "the adequate level of detail for the summary about the content used for training" and, under Article 56(9), once the act is in force the code must be ready within 9 months (perhaps by January 2025?). The IP and trade secrets of AI providers and their suppliers should be protected. Suppliers need to give AI providers the information needed to comply with the AI Act but without compromising their IP or trade secrets - Recital (88) and Article 25(5). The authorities applying the AI Act should protect IP and trade secrets - Recital (167), Article 52(6), Article 53(7), Article 78 and Annex VII, point 4.5. Adverse impact on IP rights is relevant to classifying AI systems as high risk - Recital (48). Time will tell how that plays out - recitals have a significant role in the interpretation of EU law. Conclusion: providers of GenAI must start preparing (including asking their suppliers for any information needed) and will need to think long and hard about where and how to train their models to comply with EU copyright law.

💣 The EU AI Act and IP: here are the key points. General-purpose models (GP-AI - think GenAI) are said to present "challenges to artists, authors, and other creators and the way their creative content is created, distributed, used and consumed" - Recital (105). Providers of GP-AI must have "a policy to comply with Union copyright law", including complying with opt-outs from the EU's commercial TDM exception using "state of the art technologies" - Recital (105) and Article 53(1)(c). Radically, it appears Union copyright law must be followed even where a GP-AI is trained outside of the EU: "Any provider ... should comply with this obligation, regardless of the jurisdiction in which the copyright-relevant acts underpinning the training of those general-purpose AI models take place. ... no provider should be able to gain a competitive advantage in the Union market by applying lower copyright standards than those provided in the Union." - Recital (106). Providers of GP-AI must disclose details of the content used for training - Recital (107) and Article 53(1)(d). No exception is made for open source models - Recital (104). Providers fine-tuning GP-AI need to disclose the new training data sources - Recital (109). The scope of the public disclosure is to be decided. Recital (107) says: "While taking into due account the need to protect trade secrets and confidential business information, this summary should be generally comprehensive in its scope instead of technically detailed to facilitate parties with legitimate interests, including copyright holders, to exercise and enforce their rights under Union law, for example by listing the main data collections or sets that went into training the model, such as large private or public databases or data archives, and by providing a narrative explanation about other data sources used." Under Article 56(2)(b), the AI Office needs to set out a code of practice covering "the adequate level of detail for the summary about the content used for training" and, under Article 56(9), once the act is in force the code must be ready within 9 months (perhaps by January 2025?). The IP and trade secrets of AI providers and their suppliers should be protected. Suppliers need to give AI providers the information needed to comply with the AI Act but without compromising their IP or trade secrets - Recital (88) and Article 25(5). The authorities applying the AI Act should protect IP and trade secrets - Recital (167), Article 52(6), Article 53(7), Article 78 and Annex VII, point 4.5. Adverse impact on IP rights is relevant to classifying AI systems as high risk - Recital (48). Time will tell how that plays out - recitals have a significant role in the interpretation of EU law. Conclusion: providers of GenAI must start preparing (including asking their suppliers for any information needed) and will need to think long and hard about where and how to train their models to comply with EU copyright law.

💣 The EU AI Act and IP: here are the key points. General-purpose models (GP-AI - think GenAI) are said to present "challenges to artists, authors, and other creators and the way their creative content is created, distributed, used and consumed" - Recital (105). Providers of GP-AI must have "a policy to comply with Union copyright law", including complying with opt-outs from the EU's commercial TDM exception using "state of the art technologies" - Recital (105) and Article 53(1)(c). Radically, it appears Union copyright law must be followed even where a GP-AI is trained outside of the EU: "Any provider ... should comply with this obligation, regardless of the jurisdiction in which the copyright-relevant acts underpinning the training of those general-purpose AI models take place. ... no provider should be able to gain a competitive advantage in the Union market by applying lower copyright standards than those provided in the Union." - Recital (106). Providers of GP-AI must disclose details of the content used for training - Recital (107) and Article 53(1)(d). No exception is made for open source models - Recital (104). Providers fine-tuning GP-AI need to disclose the new training data sources - Recital (109). The scope of the public disclosure is to be decided. Recital (107) says: "While taking into due account the need to protect trade secrets and confidential business information, this summary should be generally comprehensive in its scope instead of technically detailed to facilitate parties with legitimate interests, including copyright holders, to exercise and enforce their rights under Union law, for example by listing the main data collections or sets that went into training the model, such as large private or public databases or data archives, and by providing a narrative explanation about other data sources used." Under Article 56(2)(b), the AI Office needs to set out a code of practice covering "the adequate level of detail for the summary about the content used for training" and, under Article 56(9), once the act is in force the code must be ready within 9 months (perhaps by January 2025?). The IP and trade secrets of AI providers and their suppliers should be protected. Suppliers need to give AI providers the information needed to comply with the AI Act but without compromising their IP or trade secrets - Recital (88) and Article 25(5). The authorities applying the AI Act should protect IP and trade secrets - Recital (167), Article 52(6), Article 53(7), Article 78 and Annex VII, point 4.5. Adverse impact on IP rights is relevant to classifying AI systems as high risk - Recital (48). Time will tell how that plays out - recitals have a significant role in the interpretation of EU law. Conclusion: providers of GenAI must start preparing (including asking their suppliers for any information needed) and will need to think long and hard about where and how to train their models to comply with EU copyright law.

💣 The EU AI Act and IP: here are the key points. General-purpose models (GP-AI - think GenAI) are said to present "challenges to artists, authors, and other creators and the way their creative content is created, distributed, used and consumed" - Recital (105). Providers of GP-AI must have "a policy to comply with Union copyright law", including complying with opt-outs from the EU's commercial TDM exception using "state of the art technologies" - Recital (105) and Article 53(1)(c). Radically, it appears Union copyright law must be followed even where a GP-AI is trained outside of the EU: "Any provider ... should comply with this obligation, regardless of the jurisdiction in which the copyright-relevant acts underpinning the training of those general-purpose AI models take place. ... no provider should be able to gain a competitive advantage in the Union market by applying lower copyright standards than those provided in the Union." - Recital (106). Providers of GP-AI must disclose details of the content used for training - Recital (107) and Article 53(1)(d). No exception is made for open source models - Recital (104). Providers fine-tuning GP-AI need to disclose the new training data sources - Recital (109). The scope of the public disclosure is to be decided. Recital (107) says: "While taking into due account the need to protect trade secrets and confidential business information, this summary should be generally comprehensive in its scope instead of technically detailed to facilitate parties with legitimate interests, including copyright holders, to exercise and enforce their rights under Union law, for example by listing the main data collections or sets that went into training the model, such as large private or public databases or data archives, and by providing a narrative explanation about other data sources used." Under Article 56(2)(b), the AI Office needs to set out a code of practice covering "the adequate level of detail for the summary about the content used for training" and, under Article 56(9), once the act is in force the code must be ready within 9 months (perhaps by January 2025?). The IP and trade secrets of AI providers and their suppliers should be protected. Suppliers need to give AI providers the information needed to comply with the AI Act but without compromising their IP or trade secrets - Recital (88) and Article 25(5). The authorities applying the AI Act should protect IP and trade secrets - Recital (167), Article 52(6), Article 53(7), Article 78 and Annex VII, point 4.5. Adverse impact on IP rights is relevant to classifying AI systems as high risk - Recital (48). Time will tell how that plays out - recitals have a significant role in the interpretation of EU law. Conclusion: providers of GenAI must start preparing (including asking their suppliers for any information needed) and will need to think long and hard about where and how to train their models to comply with EU copyright law.

💣 The EU AI Act and IP: here are the key points. General-purpose models (GP-AI - think GenAI) are said to present "challenges to artists, authors, and other creators and the way their creative content is created, distributed, used and consumed" - Recital (105). Providers of GP-AI must have "a policy to comply with Union copyright law", including complying with opt-outs from the EU's commercial TDM exception using "state of the art technologies" - Recital (105) and Article 53(1)(c). Radically, it appears Union copyright law must be followed even where a GP-AI is trained outside of the EU: "Any provider ... should comply with this obligation, regardless of the jurisdiction in which the copyright-relevant acts underpinning the training of those general-purpose AI models take place. ... no provider should be able to gain a competitive advantage in the Union market by applying lower copyright standards than those provided in the Union." - Recital (106). Providers of GP-AI must disclose details of the content used for training - Recital (107) and Article 53(1)(d). No exception is made for open source models - Recital (104). Providers fine-tuning GP-AI need to disclose the new training data sources - Recital (109). The scope of the public disclosure is to be decided. Recital (107) says: "While taking into due account the need to protect trade secrets and confidential business information, this summary should be generally comprehensive in its scope instead of technically detailed to facilitate parties with legitimate interests, including copyright holders, to exercise and enforce their rights under Union law, for example by listing the main data collections or sets that went into training the model, such as large private or public databases or data archives, and by providing a narrative explanation about other data sources used." Under Article 56(2)(b), the AI Office needs to set out a code of practice covering "the adequate level of detail for the summary about the content used for training" and, under Article 56(9), once the act is in force the code must be ready within 9 months (perhaps by January 2025?). The IP and trade secrets of AI providers and their suppliers should be protected. Suppliers need to give AI providers the information needed to comply with the AI Act but without compromising their IP or trade secrets - Recital (88) and Article 25(5). The authorities applying the AI Act should protect IP and trade secrets - Recital (167), Article 52(6), Article 53(7), Article 78 and Annex VII, point 4.5. Adverse impact on IP rights is relevant to classifying AI systems as high risk - Recital (48). Time will tell how that plays out - recitals have a significant role in the interpretation of EU law. Conclusion: providers of GenAI must start preparing (including asking their suppliers for any information needed) and will need to think long and hard about where and how to train their models to comply with EU copyright law.