Bryan Sterba’s articles from Lowenstein Sandler have gained widespread attention:
- particularly among readers in the Technology sector
As artificial intelligence tools become more commonplace, organizations are facing a range of new challenges related to data privacy, advertising strategies, and copyright issues. In this episode of “A-I Didn’t Know That,” Bryan Sterba sheds light on how companies, especially those in regulated sectors, can responsibly incorporate AI through robust policies, training, and risk management methods. He also delves into the effects of agentic AI browsers on the ad-driven internet, along with recent developments in AI copyright litigation, including court rulings concerning training data and emerging legal challenges related to AI-generated content.
Speaker:
- Bryan Sterba, Partner, IP & Tech Transactions, Lowenstein AI, and Emerging Companies & Venture Capital
Lowenstein AI: A-I Didn’t Know That
In this series, Partner Bryan Sterba explores the complex legal and business implications of AI technology. Alongside various guests, Bryan discusses how organizations can harness AI to boost productivity and efficiency, while also addressing the myriad challenges that arise as corporations, lawmakers, and regulatory bodies navigate the diverse approaches to managing intellectual property rights, privacy, data security, and national security concerns.
READ THE TRANSCRIPT
Bryan Sterba: One of the primary insights we’ve uncovered is that a growing number of vendors are entering the AI landscape, often building products based on major foundational model providers. This creates a disconnect in terms of the safeguards surrounding client data. Many of our clients in regulated industries are apprehensive about how their data is being handled. Specifically, they ask, “How is the company I’m trusting with my data utilizing it? What about their partners, who aren’t foundational model providers themselves—how are they ensuring the protection of my data?”
Organizations understand that they must take certain risks, so crafting effective policies and educational resources has become essential. These tools help employees use AI responsibly while minimizing potential risks without completely avoiding them.
This issue is particularly significant for the Interactive Advertising Bureau and other stakeholders, as the contemporary internet relies heavily on advertising revenue. All of Google’s functionalities and many other free tools are sustained by this model.
Emerging agentic browsers have the potential to bypass the web traffic that generates advertisement views, challenging the existing digital economy that relies on such visibility. If fewer people are seeing ads, advertisers question what they are actually paying for, raising substantial legal concerns.
Interestingly, many of the major AI model providers are also significant advertisers, such as Google.
We have progressed beyond initial debates regarding training data. Recent rulings from the Northern District of California have determined that training on copyrighted materials may be considered fair use. This implies that utilizing vast amounts of copyrighted content to train foundational models is not classified as copyright infringement.
However, there has yet to be conclusive legal documentation regarding infringement resulting from the outputs of these models. While the training of models has been the central focus, concerns from authors and guilds remain. They argue that their works were appropriated to create models like ChatGPT, which could potentially replicate their content without compensation. This raises questions about recourse if the generated content closely resembles copyrighted material, given that training was deemed fair use.
Currently, these cases are in their infancy in the courts. Although they have been explored over the past few years, judges are requesting tangible examples of ChatGPT-generated content that infringes copyright before making a judgment. There is an expectation that more case law will emerge concerning potential copyright violations in this generative AI context.
Thank you for tuning in. We invite you to join us next time on “A-I Didn’t Know That.”
This article is intended to serve as a general overview of the subject. For specific advice tailored to your circumstances, please consult an expert.
