​"AI Deepfakes and the First Amendment"

_______________________________

By: Shriya Mandalapu
Tennessee
 

_______________________________

Introduction:
 

         In 2024, thousands of voters in New Hampshire received a call that used Artificial Intelligence to imitate President Joe Biden’s voice, encouraging them to not vote in the state’s primary election. The Federal Communications Commission later fined the person responsible $6 million for breaking federal law (Federal Communications Commission). Since then, thousands of AI-generated videos and audio recordings have impersonated political figures saying and doing things that never occurred. Deepfakes have grown from being an online trend we laughed at, to a threat to public trust and personal privacy. According to the US Government Accountability Office, deepfakes are AI generated videos, images, or audio recordings that make fabricated events appear real. As deepfake technology has become more accessible, it has become easier to spread misinformation on a mass scale. However, existing laws were not written with this technology in mind, and courts and lawmakers must decide how existing legal protections should apply.
 

         The greatest challenge American lawmakers face is to create laws that protect people
from the harms caused by deepfakes without violating the First Amendment, which
protects freedom of speech, press, religion, and assembly from government
interference. As artificial intelligence continues to improve, finding that balance has
become one of the most important legal issues facing the US justice system.
 

Growing Deepfake Harm:
 

         Politics has been at the greatest risk of deepfakes, as any AI influence can spread
misinformation that can impact entire countries. For example, in 2023, an election in
Slovakia is said to be greatly influenced by the spread of an AI generated recording that
falsely portrays a political candidate discussing electoral fraud (Harvard Kennedy
School). The recording spread across social media only days before voting, raising
concerns about how quickly deepfakes can travel before a fact checker can respond.
Furthermore, businesses have also become frequent targets of deepfake scams. Ferrari
CEO Benedetto Vigna was impersonated by scammers in 2024. They sent messages
and called Ferrari executives trying to fake acquisitions, and other major business
moves. The cyber attack only ceased because executives asked questions the
scammers simply could not respond to (Bloomberg Law).
 

         Deepfakes are also used to exploit public figures who are known to have a large
following. Illegal online gambling companies created AI generated videos that falsely

showed professional soccer players Jude Bellingham and Bruno Fernandes endorsing
their betting platforms without their consent (The Guardian). These fake ads mislead
customers, and also exploited the athletes' likeness for financial gain.
The harms of deepfake technology are far and wide, which only increases the pressure
on lawmakers to create legal protections against AI deepfakes.
 

Why Regulation is Difficult:

​

         Although deepfakes can cause serious harm, regulating them is not as simple as
banning false or misleading content. The First Amendment protects freedom of speech,
preventing the government from restricting speech because it is false or offensive. As a
result, lawmakers must ensure that deepfake legislation specifically addresses the
harm that deepfakes cause, instead of addressing the inherent false nature of
deepfakes. However, not all speech is protected by the First Amendment. The
government can limit speech like fraud or lies that hurt someone's reputation. However,
just because something false doesn't mean it's illegal. In United States v. Alvarez (2012),
the Supreme Court ruled that lying about military medals is protected speech unless it
causes real damage (US Courts). This makes it hard for lawmakers to ban deepfakes
just for being fake. In Brandenburg v. Ohio (1969), the Court added that speech can only
be stopped if it is meant to cause immediate violence or crime (Oyez). Since many
deepfakes just spread rumors without telling people to break the law, they are often
protected by this rule. Furthermore, the case New York Times Co. v. Sullivan (1964),
public figures seeking legal recourse against deepfakes. This decision requires
plaintiffs to prove "actual malice" (Stanford University). In other words, those attempting
to sure but need to prove that the information spread about themselves was spread
with the intent to cause reckless disregard for the truth. This protects political debate
but makes it tough for public figures to win in court. Broad prohibitions on artificial
intelligence content also risk suppressing artistic expression or satire. Effective
legislation must focus on specific dangers, such as election interference or financial
fraud. The primary challenge remains the development of safeguards against digital
harms that do not compromise fundamental rights to free expression.
 

Federal and State Responses to AI:
 

         Congress has introduced several bills to address the growing use of deepfakes, but no
federal law currently regulates all forms of AI-generated content. One of the most
significant laws is the TAKE IT DOWN Act which was signed in 2025. It requires online
platforms to remove nonconsensual explicit images, including AI-generated content,
after receiving notice from a victim (Federal Trade Commission). Congress has also

proposed the NO FAKES Act, which would prohibit the unauthorized use of a person's
voice or likeness in AI-generated media without consent (Senator Coons). Although
these proposals target some of the most harmful uses of deepfakes, they do not
establish a single legal framework for addressing all forms of AI generated media.
In the absence of federal legislation, many states have taken it upon themselves to
regulate AI deepfakes. Tennessee passed the ELVIS Act in 2024, expanding existing
right of publicity protections to cover AI-generated replicas of a person's voice
(Vanderbuilt Journal). Texas prohibits the distribution of political deepfakes intended to
influence elections before Election Day; California has also enacted laws restricting
election deepfakes. California further allows for victims of AI generated explicit images
to seek legal remedies (Ballotpedia). Although these laws provide important
protections, they differ significantly in the specific rights they grant to victims and
consequences they impose. This mismatch of regulation creates inconsistent legal
protections across the United States and highlights the need for a clear national
standard of AI regulation.
 

Conclusion:
 

         Deepfakes have become a growing challenge for the American legal system because
they can facilitate fraud. At the same time, the First Amendment protects many forms of
speech, including false statements, making the regulations of deepfake technology
challenging. As a result, lawmakers must carefully distinguish between protected
expression and harmful conduct when developing new legislation. Although Congress
and many states have begun addressing specific uses of deepfakes, current laws
remain inconsistent and leave important legal gaps. Narrowly tailored regulations that
target fraud, election interference, nonconsensual explicit images are being developed
and implemented. This narrow tailoring helps avoid potential accusations of first
amendment violations.
 

         Artificial Intelligence and deepfake technology will continue to evolve. The question is,
whether we will evolve with it.