Dazza Greenwood

On April 17, Colorado Governor Jared Polis signed HB 1058 into law, which means it is time to update our chart comparing the various definitions of sensitive data in consumer data privacy laws. Here's some more context: 1️⃣ HB 1058 amends the Colorado Privacy Act to add biological data (and through it neural data) as a category of sensitive data in the Colorado Privacy Act. 2️⃣ The true impact of the amendment is debatable as the definition of biological data states that the data must be "used or intended to be used, singly or in combination with other personal data, for identification purposes." 3️⃣ Colorado is the first state to add biological/neural data to its state privacy law. This continues the trend over the last two legislative sessions of state lawmakers expanding the types of data covered as sensitive data. Given the constant expansion, at some point it may be easier to define what is not sensitive data rather than what is. 4️⃣ California is also considering a bill to add neural data to the CCPA's definition of sensitive personal information. 5️⃣ Colorado isn't done. The legislature is still considering two other bills to amend the Colorado Privacy Act, dealing with the treatment of biometric data and minor's data. 6️⃣ Here's a link to the Colorado bill for those interested: https://lnkd.in/gE7KPyY9

134

9 Comments

Patchwork privacy protections won't shield the public from the invasive aspects of AI. Now's the time to update and bolster privacy and consumer protection laws. Don't take my word for it - Ryan Calo (University of Washington) makes quite the compelling case on this podcast from The Lawfare Institute. Give it a listen over your coffee and let me know what you think! https://lnkd.in/eXZmdvGa

9

From DC to Sacramento, AI regulation remains a hot (and often contentious) topic. Big thanks to Helen Toner and Zachary Arnold of Center for Security and Emerging Technology (CSET) for coming on The Lawfare Institute podcast to talk about a regulatory agenda that could unite "doomers" and "ethicists." Thanks also to Graham Hardig for excellent research assistance in preparing for this exciting interview. Give it a listen: https://lnkd.in/e4G3cHag

20

Jason Shull, Scott M. Kelly, and Joshua Davenport led an important discussion today on approaches to IP in an AI-driven landscape. 1 ***Define the tool to claim the work.*** We must clearly define AI’s Role in our inventions. Josh emphasized that entrepreneurs should be precise in distinguishing between AI-assisted and AI-generated outputs in their IP filings. Clear definitions can protect against legal ambiguities and help secure rights to valuable innovations. 2 ***Your work must stand apart from the machine.*** We must document our contributions and AI’s involvement. Both Scott and Josh stressed the importance of documenting the human role in the creative process – it’s crucial in proving authorship and invention in both patent and copyright applications, especially if AI was a significant part of the development process. 3 ***Secure each step along the journey*** Use multiple stages for IP protection. For products refined through AI, Josh suggested filing for protection at various stages of development. This can provide a safety net if AI contributions in the final version complicate IP claims, ensuring earlier versions remain enforceable if needed. 4 ***Novelty without inventorship is an orphaned idea.*** We must balance innovation with legal practicalities. Scott’s reflections on human inventorship highlighted that a truly novel, AI-generated idea may lack enforceable protection if human contribution is insufficient. Entrepreneurs should ensure their creative influence guides AI-generated inventions to retain ownership. 5 ***The rules are shifting as fast as the tech.*** We must stay informed on regulatory changes. Jason’s and Scott’s comments on evolving IP regulations emphasized the need for entrepreneurs to stay updated on legal guidelines around AI. Compliance with current IP laws and proactive adaptation to new policies will safeguard IP and ensure its commercial viability. Thank you Steven Zylstra and team at the Arizona Technology Council and Banner Witcoff for sponsoring and hosting, and Jason, Scott, and josh your thought-provoking insights!

14

1 Comment

California just enacted a new law governing contracts relating to “digital replicas” of performers. The law has clear implications for generative AI applications but is actually technology agnostic. A review of contracting procedures for the acquisition of “digital replica” rights is recommended for contracts that may be governed by California law. Read more about the law here.

36

2 Comments

You may have read about the recent court ruling out of Washington State that banned the use of AI-enhanced videos as evidence. Cassandre Coyer has done an excellent deep dive into what the ruling might mean for evidence, proving reliability, and black box technologies. Read more on Legaltech News | Law.com.

29

4 Comments

Ensuring our AI systems understand context and avoid unexpected outcomes is key to maintaining reliability and user satisfaction — and, in one humorous case, “Rickrolling.” My Foley colleague discusses how robust training datasets and precise AI tuning are essential to user experience and trust. #ArtificialIntelligence #AI #InnovativeTechnology #IntellectualProperty

2

Fascinating read on how alignment/misalignment, especially in language and cultural aspects lead to potential & unfortunate (unintended) biases in LLMs https://buff.ly/3Ax5XQ3 #ai #llm #bias #humancenteredAI

The NY Times and the media vs. the LLMs Interesting Article: https://lnkd.in/e-Ad8eEN Some thoughts on the article: "The dialogue surrounding the use of journalistic content by large language models (LLMs) like those developed by OpenAI, Google, Meta, and others raises critical points of contention within the AI industry—a sentiment I share with many colleagues. The core of the argument rests on the premise that without the foundational datasets provided by institutions like The New York Times, the efficacy of LLMs could be significantly compromised. Removing these data sets would necessitate a complete overhaul and retraining of these AI systems. In simple terms, it would stop LLMs dead in their tracks.   The feeling among observers of AI is that the solution for Microsoft, Google, and Meta is geared toward a resolution that, while financially demanding, acknowledges the value and necessity of these sources. Entities such as The New York Times, whose intellectual property undeniably feeds into the proficiencies of LLMs, merit fair compensation. This not only stands as a matter of ethical business practice but also poses the question—without recompense, what incentive would there be for content creators to maintain the flow of freely accessible information that others capitalize on?   However, this dispute is a visible fraction of a much larger iceberg encompassing extensive legal, societal, and technological domains. We must recognize that our current legal frameworks, including Section 230, were conceived in an era that could not have anticipated the monumental technological advances we see today. For perspective, consider that since the advent of Section 230, the power of consumer technology has amplified 3000-fold, with internet speeds accelerating by nearly 18,000 since 1996.   Section 230 was written in the era of Web 1.0 (Static Web) while we now grapple with the complexities posed by Web 2.0 (Social Web). With a broader adoption of Web 3.0 (Semantic Web)—characterized by blockchain technology with increased decentralization, greater anonymity, and the immutable nature of content—the challenges we face will only magnify. The convergence of these technological evolutions with outdated legal precedents points towards a precarious and potentially perilous path. The imperative for adaptive legal reform that keeps pace with technological innovation has never been more pressing."  

3

1 Comment

"It is without doubt that a requirement that a #patent cannot be enforced is the removal of a substantive patent right, which exceeds the statutory authority of the USPTO. Only Congress can do this, and it has not. And given the legislative history from the Senate in passing the Create Act, as a Congressional body, it had no intent to do so. Under 35 U.S.C. §2, the USPTO can only promulgate regulations that are not inconsistent with law. The NPRM is inconsistent with law, at least because there is no Congressional statute authorizing the unenforceability of a patent that has a TD that is tied directly or indirectly to a patent that has only one claim that has been held invalid, nor has that authority been unambiguously delegated to the USPTO. It is also inconsistent with law because it exceeds the ruling of Van Ornum, the only case that the USPTO cites for support." The FTC has already been sued for promulgating a rule that challengers assert exceeds the FTC's rulemaking authority. Could the USPTO's attempt to change terminal disclaimer practice trigger one or more lawsuits asserting the same type of argument? Stay tuned.

2

When AI is trained with unlicensed copyrighted material, the question of infringement versus fair use remains unresolved. Many copyright holders assert that this use violates their rights, and are litigating to address the issue. In light of this, is it premature to require AI companies to disclose which copyrighted works are used to train their generative AI systems? #ai #trainingdata

2

My most recent AI law review article is now available on SSRN: https://lnkd.in/ecx4e549 Abstract: Artificial Intelligence (AI) has transformed the technological landscape and impacted almost every field of endeavor. Predictive AI can help science and industry understand trends in vast amounts of data, create biometric identification systems, and improve business efficiency. Generative AI enables apps to create new documents, music, images, and designs. These and other forms of AI have the potential to disrupt industry, art, and culture. Although many of the concerns about AI are well known, the global and industrial opportunity is too large to pass up, and usage has become ubiquitous. This article focuses on the potential benefits and risks of various AI systems, particularly with regard to bias in the design and implementation of predictive AI, the implications to privacy by both predictive and generative AI, the consequences of inaccurate generative AI content, and the implications of adoption of generative AI if the systems prove effective. The article also addresses efforts by the European Union to adopt regulatory legislation and the U.S. alternative strategies for industry self regulation and federal legislation. Finally, the article identifies steps that industry can take to minimize risks and maximize the benefit of AI adoption.

24

My thoughts on the Senate AI Working Group's new AI Framework: “We have to wait and see what bills and roles are proposed based off of a lot of this but a lot of it is indeed promising. There's a strong emphasis on R&D, there's a strong emphasis on regulatory clarity, for example, and making sure that our existing laws are clear so that we can ensure easy, confident adoption of these technologies. There was clearly a thoughtfulness that went into this,” Mittelsteadt said. I'd estimate around 75% of the package is headed in the right direction: emphasizing engineering-based solutions, clearing up laws, and making targeted spending bets on talent and other discrete priorities. That said, elements such as AI export controls solutions in search of a problem and a recipe for policy failure. Too much is on the line to overzealously regulate. While action is needed on many fronts, with regard to export controls: “They've been there a broken record since at least 2016 about social media regulation, if not earlier. And when you cry wolf for going on 10 years about technology regulation, people stopped believing you,” said Matt Mittelsteadt, a research fellow at George Mason University’s Mercatus Center. “That's exactly where Congress is, and I think the most likely thing that we can expect is nothing."

4

"LibGen, which describes itself as a “links aggregator,” provides access to copyrighted works from publishers including Cengage Learning, Macmillan Learning, McGraw Hill, and Pearson Education. LibGen has been sued a number of times, ordered to shut down, and fined tens of millions of dollars for copyright infringement."

1

Working on bias in AI? Connect with Neeraj Chandra >>

17

1 Comment

Fascinating research highlights limitations of #LLMs and #RAG for legal research & analysis. Products from #LegalTech majors yielded incomplete or hallucinatory responses on at least 35% of the test questions. The "Typology of Legal RAG Errors" is particularly intriguing. #AI https://lnkd.in/ggV2KuQ3

21

1 Comment

Know more about the new California law that protects your neural data. #neuraldata #privacylaws #dataprotectionlaws #consumerprivacy

8

🔥 California’s AI Laws Are Changing the Game! 🔥 Imagine this: You’re building the next big AI breakthrough… and BOOM 💥 California’s new regulations hit, flipping the script on how AI giants like Google, Amazon, and Meta operate. But wait, what does this mean? Let me break it down: 📱 For Android: Privacy laws mean users now get more control over their data. Android must clean up its act and say goodbye to shady tracking. Oh, and those AI-generated features? They’ll need receipts—disclosing training data is the new normal. 🎙️ For Alexa: "Alexa, are you spying on me?" Not anymore (or at least, they’ll have to tell you). Voice data requires explicit consent. Generative AI outputs? Expect “AI-generated” labels popping up everywhere. 💬 For WhatsApp: AI-powered moderation? It better be transparent. Deepfakes on WhatsApp? Labeled or banned. You can’t just get banned by a bot anymore without knowing why. Transparency is king. 💡 The Big Picture: California’s laws are forcing AI to grow up. More transparency, better ethics, and fewer “oops” moments in AI innovation. But here’s the kicker: It’s going to cost companies big 💰, which might just make the AI giants even stronger. 🌍 Why does this matter? California’s rules might set the global standard. What starts in Silicon Valley could rewrite the playbook for AI innovation across the planet. So, is this the dawn of a more ethical AI era? Or just another speed bump on the road to tech supremacy? Drop your thoughts below 🧠👇 #AI #Regulations #CaliforniaLaws #EthicalAI #TechFuture

1

European legislators are working hard to understand generative AI, and data protection authorities are carefully assessing key legal and technical questions such as does a model itself actually contain personal information. FPF official side event at the Global Privacy Assembly will dig deep into this issue. California however.... well this bill talks about systems and of course an AI system can be trained on personal information and can output personal information. But the model itself? Maybe, may not always. The sponsor's intent is clearly intended to cover models, not only a full system. So thanks for the confusion, in the state where much of this is actually being built. I am all for advancing smart state AI bills, coordinated around a common model... not a fan of 'one off' stabs an an issue that create legal muddles.. We shared some of the emergence guidance from DPAs in Hamburg and the CNIL - but I suspect Newsom signed this bill, because he did veto the "big" AI state bill that many lobbied against. The suprise will be if this bill ends up having more impact, given the challenges it creates.

37