Artificial intelligence (AI) is impacting many facets of intellectual property law and practice. We have to keep in mind both the challenges AI presents and the opportunities it creates for innovators and legal professionals.
Copyright and AI:
The US Copyright Office has released reports examining the intersection of AI and copyright law. Their Part 2 report indicated that questions of copyright protection of AI outputs can generally be resolved under existing law. It clarified that copyright protects the original expression by a human author, even when using AI tools, but does not extend to purely AI-generated material or material with insufficient human control. The report did not find a basis for additional copyright or sui generis protection for AI-generated content.
More recently, the US Copyright Office’s Part 3: Generative AI Training report concluded that many current industry practices for AI training probably do not qualify as fair use. This report, released in pre-publication form, suggests that uploading entire copyrighted works for training is prima facie infringement unless licensed or excused by fair use. This has intensified and politicized the debate over compensating creators for AI training data. The recent turmoil at the US Copyright Office, including the firing of the Director of the Copyright Office, has occurred alongside these reports and raises questions about the future approach to AI, although the report is expected to remain influential.
Understanding copyright risks related to generative AI is increasingly critical for businesses. For example, Warner Music Group sued DSW for using copyrighted songs in social media ads without permission, potentially leading to significant damages.
Legislative bodies are considering the impact of AI on copyright. The Hong Kong government consulted on potential revisions to its Copyright Ordinance in view of generative AI developments. In the UK, the Copyright Licensing Agency is developing a Generative AI Training Licence, and UK artists have urged Parliament to protect copyright and require AI companies to disclose copyrighted works used for training.
Courts are grappling with AI-related copyright issues. In India, the Delhi High Court is poised to rule on the lawfulness of Text and Data Mining (TDM) for machine learning, considering whether human learning equates to machine learning. Chinese courts have seen an increase in AI-related disputes, including a ruling on AI voice infringement. While one Beijing Court ruling upheld copyright for an AI work with human originality, a more recent source indicates a China court has ruled for the first time that AI-generated content is not capable of copyright protection.
There are ongoing discussions about collective agreements on AI in film and TV production and concerns about unfair licensing practices for digital resources potentially used for TDM by libraries.
Patents and AI:
The patentability of AI-related inventions remains a key issue. The Federal Circuit’s Recentive Analytics v. Fox Corp. decision was the first to explicitly address patent eligibility in the context of AI use. While seemingly challenging for AI patents, strategies for successful drafting and prosecution still exist despite increased USPTO scrutiny. The Patent Trial and Appeal Board (PTAB) has also denied patents for AI-based medical tools based on eligibility, rather than novelty.
AI is increasingly being explored and implemented in patent practice. There are AI tools for improving invention harvesting and drafting which are critical tasks both for in-house counsel and external advisors. Companies are developing and offering AI-powered tools for various patent workflows, including ideation, patentability evaluation, search, and analytics. Examples include IP.com’s IQ Ideas+ and InnovationQ+, Clarivate’s patent intelligence solutions and Unified Patents’ AI landscape and claim charting tools. The use of responsible AI for patent drafting by law firms is also a key issue.
Discussions around AI are prompting broader questions about the patent system. Influential figures such as Jack Dorsey and Elon Musk have questioned the need for IP law in the age of AI. The AIPLA has highlighted concerns about patent eligibility and AI/IP to the US administration. The Japan Patent Office has issued examination guidance on AI-related technologies. China holds a significant percentage of granted generative AI patents, and Chinese companies are focusing on commercialising their patent assets.
Trademarks and AI:
The emerging area of AI-generated trademarks presents challenges regarding legal ownership of newly created brands (although this may be better characterised as a copyright issue in many jurisdictions).
AI is seen as a tool that can both create brand protection issues and help combat fakes and brand impersonation. Companies like Red Points and Corsearch offer brand protection solutions utilizing AI.
The use of AI by trademark practitioners and IP offices is being explored. An amended Trademark Law is reportedly in the works in China.
Trade Secrets and AI:
Generative AI introduces risks of unintentional disclosure of trade secrets. We should expect that courts are likely to protect trade secret owners against new methods of theft facilitated by technology, including generative AI. However, it does require additional diligence to ensure that trade secrets don’t inadvertently end up in a large language models and thereafter slip out again. UK businesses using generative AI need to implement “reasonable steps,” such as robust access controls, to comply with trade secret laws and mitigate disclosure risks.
New contractual safeguards such as AI circuit breakers are being considered in IT contracts for AI systems to prevent undesirable behavior that could lead to harm or disclosure. Recent trade secret litigation has demonstrated the significant remedies available for misappropriation, particularly involving former employees and competitors.
The intersection of AI and IP is continuing to evolve at a frenetic pace, with significant legal and practical implications unfolding rapidly across copyright (especially regarding training data and authorship), patents (for AI inventions and in prosecution tools), trademarks (in protection and generation), and trade secrets (in risks and safeguards). Policy discussions and legislative initiatives are underway in various jurisdictions to attempt to address these challenges and opportunities.
(The first draft of this post was created with help from NotebookLM.)