Blog

March 6, 2024 Artificial Intelligence, Ethics

Lawyers and AI: Key takeaways from being on a panel at a legal ethics conference

Earlier today I was on a panel at Hinshaw & Culbertson’s LMRM Conference in Chicago. This was the 23rd annual LMRM Conference, and the event has become the gold standard for events that focus on the “law of lawyering.”

Our session was titled How Soon is Now—Generative AI, How It Works, How to Use it Now, How to Use it Ethically. Preparing for and participating in the event gave me the opportunity to seriously consider some of the key issues relating to how lawyers are using generative AI and the promise that wider future adoption of these technologies in the legal industry holds.

Here are a few key takeaways:

    • Effective use. Lawyers are already using generative AI in ways that aid efficiency. The technology can summarize complex texts during legal research, allowing the attorney to quickly assess if the content addresses her specific interests, is factually relevant, and aligns with desired legal outcomes. With a carefully crafted and detailed prompt, an attorney can generate a pretty good first draft of many types of correspondence (e.g., cease and desist letters). Tools such as ChatGPT can aid in brainstorming by generating a variety of ideas on a given topic, helping lawyers consider possible outcomes in a situation.

 

    • Access to justice. It is not clear how generative AI adoption will affect access to justice. While it is possible that something like “legal chatbots” could bring formerly unavailable legal help to parties without sufficient resources to hire expensive lawyers, the building and adoption of sophisticated tools by the most elite firms will come at a cost that is passed on to clients, making premium services even more expensive, thereby increasing the divide that already exists.

 

    • Confidentiality and privacy. Care must be taken to reduce the risk of unauthorized disclosure of information when law firms adopt generative AI tools. Data privacy concerns arise regardless of the industry in which generative AI is used. But lawyers have the additional obligation to preserve their clients’ confidential information in accordance with the rules governing the attorney-client relationship. This duty of confidentiality complicates the ways in which a law firm’s “enterprise knowledge” can be used to train a large language model. And lawyers must consider whether and how to let their clients know that the client’s information may be used to train the model.

 

    • Exposing lawyering problems. Cases such as Mata v. Avianca, Park v. Kim and Kruse v. Karlen—wherein lawyers or litigants used AI to generate documents submitted to the court containing non-existent case citations (hallucinations)—tend to be used to critique these kinds of tools and tend to discourage lawyers from adopting them. But if one looks at these cases carefully, it is apparent that the problem is not so much with the technology, but instead with lawyering that lacks the appropriate competence and diligence.

      •  

    • AI and the standard of the practice. There is plenty of data suggesting that most knowledge work jobs will be drastically impacted by the use of AI in the near term. Regardless of whether a lawyer or law firm wants to adopt generative AI in the practice of law, attorneys will not be able to avoid knowing how the use of AI will change norms and expectations, because clients will be effectively using these technologies and innovating in the space.

Thank you to Barry MacEntee for inviting me to be on his panel. Barry, you did an exemplary job of preparation and execution, which is exactly how you roll. Great to meet my co-panelist Andrew Sutton. Andrew, your insights and commentary on both the legal and technical aspects of the use of AI in the practice of law were terrific.

March 6, 2024 Computer Crime, Trade Secrets

Fifth Amendment did not save former employee from having to turn over his Gmail account

Gmail Fifth Amendment

Plaintiff biotech company sued a former employee for allegedly emailing proprietary information to his personal Gmail account and discussing employment with competitors. Plaintiff’s investigations revealed defendant had sent over a hundred emails with confidential data to his Gmail account, in violation of a confidentiality agreement defendant had signed when he was hired. Plaintiff sued defendant alleging misappropriation of trade secrets under both federal and state law. Plaintiff sought a temporary restraining order that required defendant to turn over his devices and online accounts for inspection. The court granted the motion.

Injunctive relief warranted

The court found that plaintiff had shown a reasonable probability of success in the litigation. It had successfully alleged ownership of trade secrets and had described specific instances (e.g., sending emails to a private Gmail account) that would be considered misappropriation.

Defendant could not be trusted

As for the likelihood of irreparable harm plaintiff would suffer if the injunction were not granted, the court considered plaintiff’s assertion that defendant “could not be trusted” based on his alleged conduct, and that plaintiff would suffer irreparable harm because of the continued presence of unsecured confidential information on defendant’s devices and accounts.

No Fifth Amendment Protection

Defendant argued under the “balancing of the equities” test that requiring him to turn over his devices and accounts would violate his Fifth Amendment rights against self-incrimination. The court rejected this argument, however, observing that in the course of plaintiff’s investigation of defendant’s conduct, defendant signed a document knowingly, intelligently and voluntarily, thereby admitting there was incriminating evidence to be found. Because of this, the court found defendant waived his Fifth Amendment rights.

 Injunction favored the public interest

The court also found that entry of the injunction requiring defendant to turn over the devices and accounts would benefit the public interest. It noted that there is a generalized public interest in upholding the inviolability of trade secrets and enforceability of confidentiality agreements. It mentioned the general interest in preserving Fifth Amendment rights but reiterated that in these circumstances, because of defendant’s waiver, the Fifth Amendment did not shield defendant.

Legend Biotech USA v. Liu, 2024 WL 919082 (D.N.J. March 4, 2024)

See also:

March 5, 2024 Privacy

Website operator not liable under Wiretap Act for allowing Meta to intercept visitor communications

Plaintiffs asserted that defendant healthcare organization inadequately protected the personal and health information of visitors to defendant’s website. In particular, plaintiffs alleged that unauthorized third parties – including Meta – could intercept user interactions through the use of tracking technologies such as the Meta Pixel and Conversions API. According to plaintiffs, these tools collected sensitive health information and sent it to Meta. Despite defendant’s privacy policy claiming to protect user privacy and information, plaintiffs alleged that using defendant’s website caused plaintiffs to receive unsolicited advertisements on their Facebook accounts.

Plaintiffs sued, asserting a number of claims, including under the federal Electronic Communications Privacy Act (“ECPA”) and the California Invasion of Privacy Act (“CIPA”). Defendant moved to dismiss these claims. The court granted the motion.

To establish an ECPA claim, a plaintiff must demonstrate that defendant intentionally intercepted or attempted to intercept electronic communications using a device. CIPA similarly prohibits using electronic means to understand the contents of a communication without consent. Both laws have a “party exception” allowing a person who is a party to the communication to intercept it, provided the interception is not for a criminal or tortious purpose. In other words, there is an exception to the exception.

In this case, defendant argued it was a legitimate party to plaintiffs’ communications on a website, thus invoking the party exception. Plaintiffs countered that the exception should not apply due to defendant’s alleged tortious intent (making the information available to Facebook without disclosure to plaintiffs). But the court found that plaintiffs did not provide sufficient evidence that defendant’s actions were for an illegal or actionable purpose beyond the act of interception itself. Under the guidance of Pena v. GameStop, Inc., 2023 WL 3170047 (S.D. Cal. April 27, 2023), (a plaintiff must plead sufficient facts to support an inference that the offender intercepted the communication for the purpose of a tortious or criminal act that is independent of the intentional act of recording or interception itself), the court concluded there was no separate tortious conduct involved in the interception and dismissed the claims.

B.K. v. Eisenhower Medical Center, 2024 WL 878100 (February 29, 2024)

See also:

March 4, 2024 Artificial Intelligence

VIDEO: Elon Musk / OpenAI lawsuit – What’s it all about?

 

So Elon Musk has sued OpenAI. What’s this all about?

The lawsuit centers on the breach of a founding agreement and OpenAI’s shift from non-profit to for-profit through partnerships with companies like Microsoft. It has been filed in state court in California and talks about the risks of artificial general intelligence (or AGI). It talks about how Musk worked with Sam Altman back in 2015 to form OpenAI for the public good. That was the so called “founding agreement” which also got written into the company’s certificate of incorporation. One of the most intriguing things about the lawsuit is that Musk is asking the court to determine that OpenAI is Artificial General Intelligence and thereby has gone outside the initial scope of the Founding Agreement.

March 4, 2024 Computer Crime

Stealing data: Ninth Circuit examines whether cellular data can be subject to a conversion claim

data conversion

Plaintiffs sued Google alleging Google improperly used plaintiffs’ cellular data without consent, constituting conversion under California law. The lower court dismissed the case for failure to state a claim. Plaintiffs sought review with the Ninth Circuit. On appeal, the court reversed the lower court’s decision concerning the conversion claim, finding that cellular data is something that can be subject to conversion.

The court observed that a successful conversion plaintiff must plead and prove (1) ownership or rightful possession of property, (2) defendant’s use of the property in violation of plaintiff’s rights, and (3) resulting damages. The court found that plaintiffs satisfactorily established cellular data as a form of personal property subject to conversion, given its definable nature, potential for exclusive control, and plaintiffs’ legitimate expectations based on their data plans.

Moreover, the court concluded that plaintiffs’ allegations against Google meet the criteria for conversion, demonstrating unauthorized use of their cellular data that went against their property interests and resulted in quantifiable damages. By equating Google’s actions to a “forced sale” of plaintiffs’ data, the court underscored the tangible impact of intangible property loss.

Taylor v. Google, 2024 WL 837044 (9th Cir. February 28, 2024)

See also:

February 29, 2024 Contracts, Litigation

Nvidia forces consumer lawsuit into arbitration  

arbitration provisoin

Plaintiffs filed a class action suit against Nvidia alleging that Nvidia falsely advertised a game streaming feature for its Shield line of devices which was later disabled, thus depriving consumers of a paid feature and devaluing their devices. The suit included claims of trespass to chattels, breach of implied warranty, and violations of various consumer protection laws.

Nvidia filed a motion to compel arbitration, citing an agreement that users ostensibly accepted during the device setup process. This agreement provided that disputes would be resolved through binding arbitration in accordance with Delaware laws and that any arbitration would be conducted by an arbitrator in California.

The court looked to the Federal Arbitration Act, which upholds arbitration agreements unless general contract defenses like fraud or unconscionability apply. Nvidia emphasized the initial setup process for Shield devices, during which users were required to agree to certain terms of use that included the arbitration provision. In light of Nvidia’s claim that this constituted clear consent to arbitrate disputes, the court examined whether this agreement was conscionable and whether it indeed covered the plaintiffs’ claims.

The court found the arbitration agreement enforceable, rejecting plaintiffs’ claims of both procedural and substantive unconscionability. The court concluded that the setup process provided sufficient notice to users about the arbitration agreement, and the terms of the agreement were not so one-sided as to be deemed unconscionable. Furthermore, the court determined that plaintiffs’ claims fell within the scope of the arbitration agreement, leading to a decision to stay the action pending arbitration in accordance with the agreement’s terms.

Davenport v. Nvidia Corporation, — F.Supp.3d —, 2024 WL 832387 (N.D. Cal. Feb 28, 2024)

See also:

February 23, 2024 Artificial Intelligence, Litigation

ChatGPT was “utterly and unusually unpersuasive” in case involving recovery of attorney’s fees

chatgpt billing

In a recent federal case in New York under the Individuals with Disabilities Act, plaintiff prevailed on her claims and sought an award of attorney’s fees under the statute. Though the court ended up awarding plaintiff’s attorneys some of their requested fees, the court lambasted counsel in the process for using information obtained from ChatGPT to support the claim of the attorneys’ hourly rates.

Plaintiff’s firm used ChatGPT-4 as a “cross-check” against other sources in confirming what should be a reasonably hourly rate for the attorneys on the case. The court found this reliance on ChatGPT-4 to be “utterly and unusually unpersuasive” for determining reasonable billing rates for legal services. The court criticized the firm’s use of ChatGPT-4 for not adequately considering the complexity and specificity required in legal billing, especially given the tool’s inability to discern between real and fictitious legal citations, as demonstrated in recent past cases within the Second Circuit.

In Mata v. Avianca, Inc., 2023 WL 4114965 (S.D.N.Y. June 22, 2023) the district court judge sanctioned lawyers for submitting fictitious judicial opinions generated by ChatGPT, and in Park v. Kim, — F.4th —, 2024 WL 332478 (2d Cir. January 30, 2024) an attorney was referred to the Circuit’s Grievance Panel for citing non-existent authority from ChatGPT in a brief. These examples highlighted the tool’s limitations in legal contexts, particularly its inability to differentiate between real and fabricated legal citations, raising concerns about its reliability and appropriateness for legal tasks.

J.G. v. New York City Dept. of Education, 2024 WL 728626 (February 22, 2024)

See also:

February 22, 2024 Section 230

Section 230 protects Snapchat against lawsuit brought by assault victim

section 230

A young girl named C.O. found much misfortune using Snapchat. Her parents (the plaintiffs in this lawsuit) alleged that the app’s features caused her to become addicted to the app, to be exposed to sexual content, and to eventually be victimized on two occasions, including once by a registered sex offender.

Suing Snapchat

Plaintiffs sued Snap and related entities, asserting claims including strict product liability, negligence, and invasion of privacy, emphasizing the platform’s failure to protect minors and address reported abuses. Defendants moved to strike the complaint.

The court granted the motion to strike. It held that the allegations of the complaint fell squarely within the ambit of immunity afforded under Section 230 to “an interactive computer service” that acts as as a “publisher or speaker” of information provided by another “information content provider.” Plaintiffs “clearly allege[d] that the defendants failed to regulate content provided by third parties” when such third parties used Snapchat to harm plaintiff.

Publisher or speaker? How about those algorithms!

Plaintiffs had argued that their claims did not seek to treat defendants as publishers or speakers, and therefore Section 230 immunity did not apply. Instead, plaintiffs argued, they were asserting claims that defendants breached their duty as manufacturers to design a reasonably safe product.

Of particular interest was the plaintiffs’ claim concerning Snapchat’s algorithms which recommended connections and which allegedly caused children to become addicted. But in line with the case of Force v. Facebook, Inc., 934 F.3d 53 (2nd Cir. 2019), the court refused to find that use of algorithms in this way was outside the traditional role of a publisher. It was careful to distinguish the case from Lemmon v. Snap, Inc., 995 F.3d 1085 (9th Cir 2021), in which that court held Section 230 did not immunize Snapchat from products liability claims. In that case, the harm to plaintiffs did not result from third party content but rather from the design of the platform which tempted the users to drive fast. In this case, the harm to plaintiffs was the result of particular actions of third parties who had transmitted content using Snapchat, to lure C.O.

Sad facts, sad result

The court seemed to express some trepidation about its result, using the same language the First Circuit Court of Appeals used in Jane Doe No. 1 v. Backpage.com, LLC, 817 F.3d 12, 15 (1st Cir. 2016): “This is a hard case-hard not in the sense that the legal issues defy resolution, but hard in the sense that the law requires that [the court] … deny relief to plaintiffs whose circumstances evoke outrage.” And citing from Vazquez v. Buhl, 90 A.3d 331 (2014), the court observed that “[w]ithout further legislative action, however, there is little [this court can] do in [its] limited role but join with other courts and commentators in expressing [its] concern with the statute’s broad scope.”

V.V. v. Meta Platforms, Inc. et al., 2024 WL 678248 (Conn. Super. Ct., February 16, 2024)

See also:

February 21, 2024 Copyright

Fourth Circuit overturns massive jury verdict in copyright case against internet service provider

music infringement

Plaintiff copyright holders sued defendant internet service provider alleging both vicarious and contributory copyright infringement liability arising from defendant’s customers downloading or distributing songs using BitTorrent. The jury found defendant liable and awarded $1 billion in statutory damages. Defendant sought review with the Fourth Circuit. On appeal, the court affirmed the jury’s finding of willful contributory infringement but remanded the action for a new trial on damages because it found plaintiffs failed to prove vicarious liability, as defendant did not profit from its subscribers’ acts of infringement.

No vicarious liability

Citing to Metro-Goldwyn-Mayer Studios Inc. v. Grokster, Ltd., 545 U.S. 913 (2005) and CoStar Grp., Inc. v. LoopNet, Inc., 373 F.3d 544 (4th Cir. 2004), the court observed that “[a] defendant may be held vicariously liable for a third party’s copyright infringement if the defendant ‘[1] profits directly from the infringement and [2] has a right and ability to supervise the direct infringer.’” In this case, the court found that plaintiffs failed to prove that defendant profited directly from its subscribers’ copyright infringement.

The crux of the financial benefit inquiry was whether a causal relationship existed between the subscribers’ infringing activity and defendant’s financial benefit. To prove vicarious liability, plaintiff had to show that defendant profited from its subscribers’ infringing download and distribution of plaintiffs’ copyrighted songs. The court found that plaintiffs did not meet that burden.

The appellate court disagreed with the lower court’s determination that defendant’s repeated refusal to terminate infringing subscribers’ accounts was enough to show financial benefit for these purposes. Instead, the court found that continued payment of monthly fees for internet service, even by repeat infringers, was not a financial benefit flowing directly from the copyright infringement itself. “Indeed, Cox would receive the same monthly fees even if all of its subscribers stopped infringing.”

The court rejected plaintiffs’ alternative theories for financial benefit. Plaintiffs argued that the high volume of infringing activity on defendant’s network, with roughly 13% of traffic from peer-to-peer activity and over 99% of that being infringing, suggested that the ability to infringe attracted customers to defendant’s internet service. However, the evidence did not conclusively show that customers chose defendant’s service specifically for its potential to facilitate copyright infringement. The argument overlooked the fact that internet service is essential for many aspects of modern life, and there was no specific evidence that defendant’s internet service was selected over competitors due to a more lenient stance on copyright infringement.

Additionally, plaintiffs claimed that defendant’s subscribers were willing to pay more for internet services that allowed for copyright infringement, citing defendant’s tiered pricing and the correlation between peer-to-peer activity and higher data usage. However, there was no substantial evidence to support the claim that subscribers chose higher internet speeds with the intention of infringing copyright. Plaintiffs’ own expert acknowledged that increased data usage could be attributed to numerous legal activities like streaming and gaming. The argument failed to establish a direct link between the desire for higher internet speeds and the intent to infringe copyright, leaving plaintiffs’ assertion that defendant profited from copyright infringement unsubstantiated. Consequently, the court found no basis for vicarious liability on defendant’s part for its subscribers’ copyright infringements, making it necessary to overturn the lower court’s decision on this issue.

Contributory liability upheld

The court upheld the lower court’s determination that defendant was contributorily liable for its subscribers’ infringement, finding that defendant was aware of and materially contributed to the infringing activities. The court emphasized the need for defendant to have knowledge of specific instances of infringement and the substantial certainty of continued infringement by particular subscribers. Despite defendant’s tiered internet services and a variety of lawful uses, the evidence presented at trial demonstrated defendant’s knowledge of repeat infringements and its decision to continue providing service to infringing subscribers, primarily to avoid losing revenue. The court rejected defendant’s arguments against contributory liability, affirming that providing a service with knowledge of its use for infringement, especially when specific instances are known, constituted material contribution to infringement.

But what are the damages now?

Because the $1 billion damages award was not allocated between the two theories of liability, and the jury was instructed to consider various factors, including the profits defendant earned from the infringements, the court could be sure that the vicarious liability verdict did not impact the damages awarded. Given this uncertainty and the significant discretion granted to the jury in determining statutory damages, the court vacated the damages award and remanded for a new trial on the damages issue.

Sony Music Entertainment v. Cox Communications, Inc., 2024 WL 676432 (4th Cir., February 20, 2024)

See also:

Scroll to top