Samuelson-Glushko Technology Law & Policy Clinic (TLPC)

By Neven Grigic, Student Attorney

On March 31, the Clinic filed an amicus brief on behalf of three youth organizations in the first case of its kind to consider whether the outputs of generative AI systems are entitled to First Amendment protections. Our brief on behalf of Encode, Design It For Us, and Young People’s Alliance was filed in the Middle District of Florida in the tragic case of Garcia v. Character Technologies, where the mother of a teenage boy who took his own life following a series of emotionally abusive interactions with an AI chatbot has brought suit against its makers for tortious conduct and violations of products liability laws.

Student attorneys Santana Andazola, Jordan Chen, Fynn Fehrenbach, Natalie Phillips, and Neven Grigic drafted the brief under TLPC Director Vivek Krishnamurthy’s supervision. Zoe Leonore Glepa and Telly Scott handled cite-checking and final revisions, while Heidi Mehaffey of Mehaffey P.A. served as our local Florida counsel.

Background

Character Technologies is the developer of an AI chatbot called “Character AI,” which allows users to chat with AI-generated characters—including personas drawn from works of fiction. Plaintiff Megan Garcia seeks to hold Character Technology (and several others) liable for flaws in the design of Character AI, which exposed her 14-year-old son to sexual content that led to addiction, depression, and anxiety. Ms. Garcia alleges that these interactions, including alleged incitements by the chatbot, led her son to end his own life.

Character AI and the other defendants have filed motions to dismiss—a legal tool used to end proceedings before fact-finding when a plaintiff cannot succeed even if all alleged facts are true. The defendants further argue the outputs of its chatbot constitute protected “speech” under the First Amendment, which bars the plaintiff’s claims. Additionally, they contend that users’ First Amendment rights to receive C.AI content would be violated if the plaintiff succeeds.

Our Arguments

On behalf of our clients, TLPC’s amicus brief urged the Court not to dismiss this case of first impression at this early stage of the proceedings, before there has been an opportunity for the development of a factual record. Furthermore, our amicus brief presented several arguments rebutting Character Technologies’ claims that the outputs of its chatbots are protected speech under the First Amendment. Specifically, our amicus brief argues:

1. LLM outputs should not qualify as “speech” under the First Amendment because they lack human intent and expressive purpose, instead generating text through statistical methods applied to training data.
2. Just as copyright law denies protection to non-human generated content, the First Amendment should not extend to LLM outputs. This extension would prevent courts and legislatures from addressing LLM-specific harms.
3. First Amendment issues require factual inquiry that cannot be completed at the motion to dismiss stage. Dismissal without a developed factual record would be premature.

Judge Anne Conway will hear arguments on the defendants’ motion to dismiss Ms. Garcia’s lawsuit on April 28, 2025 at the federal courthouse in Orlando, Florida. TLPC’s amicus brief is available to read and download from the link below.

(By Benjamin Yamada, Student Attorney, TLPC)

On behalf of the Wikimedia Foundation and a diverse international coalition of clients, the TLPC recently filed an amicus brief urging the California Supreme Court to reject a controversial new reading of the Stored Communications Act (“SCA”) that significantly weakens online privacy protections. Student attorneys Fynn Fehrenbach, Telly Scott, and Benjamin Yamada led the development and drafting of the brief, supervised by Vivek Krishnamurthy, TLPC’s Director, and by Vanessa Racehorse on the CU Law Faculty. The other signatories of the brief drafted by TLPC are Bolo Bhi (Pakistan), Digital Rights Foundation (Pakistan) Open MIC (USA), Software Freedom Law Center (India), and Tech Global Institute (Canada).

The Case

In Snap, Inc. v. Superior Court (Pina), defendant Adrian Pina—accused of murdering his brother—subpoenaed Snap (Snapchat) and Meta (Facebook, Instagram) to disclose the victim’s private account contents. Pina’s defense counsel hoped to uncover potentially violent or aggressive posts that might support a self-defense claim.

Snap and Meta refused, invoking the SCA’s “privacy wall,” which typically bars providers from sharing users’ stored messages, photos, and other account data unless an explicit legal exception applies (like a warrant). The California Court of Appeals, however, devised a novel “business purpose” interpretation of the SCA. Under this approach, if a provider can access user data for any internal business reason—beyond basic storage and processing—the SCA’s privacy protections do not apply. The court concluded that platforms like Snapchat, Instagram, and Facebook presumably retain such a right of access; hence, § 2702(a)’s disclosure ban does not shield those accounts.

Why the Decision Matters

Until now, § 2702 has been understood as the critical backbone of user privacy online—both in the U.S. and abroad. Email providers, social networks, and cloud storage services rely on it to refuse third-party requests (or demands) to share user messages, photos, and more. By effectively discarding that shield whenever companies have a “business purpose” for stored content, the court’s ruling undermines years of established SCA jurisprudence. For users living under authoritarian regimes who depend on U.S.-based platforms to keep them safe from government persecution, this ruling creates a major vulnerability. Foreign authorities could use the “business purpose” logic to demand user data without going through proper judicial channels.

Our Argument

Our primary concern lies with preserving robust user privacy protections for everyone. We believe the Court of Appeal’s reasoning erodes crucial safeguards for Internet users everywhere. Specifically, we argue that Section 2702 has long been interpreted by users and lawmakers to be a “blocking statute” that prevents disclosure as a default, with only narrow exceptions. The new reading upends this consensus. Additionally, we argue that the “business purpose” theory proposed by the lower court overlooks the mutual legal assistance mechanisms whereby upright foreign governments can bypass the SCA’s robust blocking function. The “business purpose” theory renders these mechanisms moot. 

The unified stance of our amici is clear: the Court of Appeal’s novel “business purpose” rationale breaks from longstanding SCA precedent and threatens to expose countless Internet users—including journalists, activists, and vulnerable communities—to unwarranted disclosures.

Our amicus brief is available below.

By Farieha Aziz (Co-Founder, Bolo Bhi) & Vivek Krishnamurthy (Director, TLPC)

Since the beginning of 2024, network disruptions and slow internet speeds have become the norm in Pakistan. Various reasons have been cited, including faults in undersea cables, upgrades to a web-management system, and the installation of a national firewall—sometimes referred to interchangeably.

Government officials and regulators have provided contradictory accounts, neither fully acknowledging the deployments nor clarifying their exact nature. Nevertheless, their narrative consistently emphasizes that firewalls exist in many countries and that content regulation, including restrictions on speech about state institutions and public officials, is a global norm. The justification for these measures is typically framed in terms of national security, counterterrorism, and the fight against “fake news.”

Earlier in 2024, Bolo Bhi and the Samuelson-Glushko Technology Law and Policy Clinic (TLPC) at the University of Colorado Law School worked together to examine Pakistan’s approach to regulating disinformation. This included analyzing changes to Pakistan’s laws in comparison with international approaches and standards.

This term, TLPC students Jordan Chen, Santana Andazola, Neven Grigic, and Natalie Phillips explored content regulation and speech restrictions in the context of claims that such practices are common worldwide—even in the US and EU. The project explored whether Pakistan’s restrictions— implemented through both technological measures and legislative changes—align with those in other jurisdictions. The resulting memoranda, which are compiled in this document, provide insights into what actually happens in other regions and highlight how these practices differ from Pakistan’s approach.

TLPC is honored to share the important work our clinic has been doing to help improve access to communications for Americans who are deaf or hard of hearing.

In May 2024, a TLPC team led by student attorneys Sebastian Blitt, Madeline Finlayson, Sarah Misché, Kevin Nguyen, and Sophie Pickering filed a petition calling on the FCC to re-evaluate its approach to the use of automated speech recognition (ASR) to generate captions for phone calls. ASR allows services like Apple’s Siri or Amazon’s Alexa to recognize a user’s voice and transcribe it to text, but as users of these technologies know, they often make mistakes.

In 2018, the FCC authorized providers of Internet Protocol Caption Telephone Services (IP CTS) to rely exclusively on ASR to caption telephone calls for persons who are deaf and hard of hearing. Our petition, on behalf of three advocacy groups—TDIforAccess, the National Association of the Deaf, and the Hearing Loss Association of America—urges the FCC to change course and provide IP CTS users with the ability to seek human assistance when ASR malfunctions.

Although ASR technology is improving, the petition prepared by TLPC highlights how ASR performs poorly in recognizing the speech of persons with non-standard accents and speech patterns. Such individuals tend to be on the margins of American society and are more likely than most to live and work in noisy environments—which further degrades the performance of ASR. To meet the legal requirement of functional equivalence, the TLPC petition urges the FCC to require IP CTS providers to make a human communication assistant available whenever one is needed to communicate effectively.

Following a public comment period launched by the FCC in August, last month TLPC student attorneys Sarah Baldwin, Zoe Leonore Glepa, and Victor Laudano spearheaded the submission of a reply comment to the FCC. The reply comment highlighted how the proposed rule would drive innovation in the IP CTS marketplace and help the FCC develop performance metrics for these services—which it has failed to do for nearly a decade.

The full text of the petition and the reply comments prepared by TLPC on behalf of its clients are available below. We look forward to the FCC’s decision on our petition in the coming months.