Deepfake technology was initially developed for film production but has rapidly become a tool for criminals. It is now used for image forgery, spreading false information, fraud, and exploitation, posing significant risks to personal privacy and public trust. These AI-generated fake content can severely infringe on corporate or individual privacy and even disrupt democratic processes by spreading fake news or inciting violence.
As Deepfake technology continues to evolve, the associated threats grow. Current laws are insufficient to prevent misuse, so nations must actively legislate to address these urgent challenges. At the same time, the involvement of businesses and civil society is crucial to promoting transparent, responsible, and ethical AI development, creating a safe and trustworthy digital environment.
Global Regulations on Deepfake Technology
China
In November 2022, China issued the “Regulations on the Management of Deep Synthesis of Internet Information Services”, becoming one of the few countries specifically regulating Deepfake technology. The regulation mandates the disclosure of Deepfake content. It requires any content using this technology to be marked as such, effectively combating misinformation and protecting the public from manipulated media. This regulation, which took effect in January 2023, covers almost the entire lifecycle of Deepfake technology, from development to dissemination, with strict regulations on providers of Deepfake technology to ensure transparency and accountability.
European Union
The European Union has incorporated the regulation of Deepfake technology into its broader 2030 Digital Policy Framework. This “Digital Decade” policy aims to “empower businesses and individuals to co-create a human-centered, sustainable, and prosperous digital future,” with cybersecurity listed as one of six key focuses. The EU’s regulations related to Deepfake include:
- The AI Regulatory Framework
- The General Data Protection Regulation
- The Copyright Regime
- The e-Commerce Directive
- The Digital Services Act
- The Audio Visual Media Directive
- The Code of Practice on Disinformation
- The Action Plan on Disinformation
- The Democracy Action Plan
The AI Regulatory Framework provides specific guidelines for developing and using AI technologies, with the Digital Services Act further mandating that digital platforms monitor and manage Deepfake content. Additionally, the EU has established the Code of Practice on Disinformation to counteract Deepfake and other forms of misleading content.
South Korea
South Korea’s regulatory approach combines legal measures with technological innovation to counter Deepfake misuse, particularly in protecting individual rights like copyright and preventing the societal risks posed by misinformation. Building on the “Digital Rights Bill” enacted last September, the South Korean government announced a series of legislative plans in May to reform the current copyright regime in response to AI-generated content and Deepfake misuse leading to the spread of fake news. Meanwhile, the South Korean government continues to invest in AI research, including developing advanced tools to detect and manage Deepfake technology.
United Kingdom
In April of this year, the UK government announced a new legislative proposal to criminalize the creation of sexual Deepfake images, introduced as an amendment to the “Criminal Justice Bill” passed in June. Initially, the “Online Safety Act” (OSA) classified the distribution of sexual Deepfake images as a criminal offense.
Under the new bill, anyone using “computer images or any other digital technology” to create or design another person’s sexual images to cause distress, anxiety, or humiliation may face a criminal record and unlimited fines. If the offender intends to share or distribute the sexual images, they could be prosecuted and face up to 2 years in prison.
United States
The United States currently lacks federal legislation specifically targeting Deepfake technology, but related regulations are already well-established. The “Identifying Outputs of Generative Adversarial Networks Act” requires the National Science Foundation to support research into standards for detecting and identifying GAN (Generative Adversarial Networks) outputs. The “Deepfake Report Act of 2019” and the “DEEPFAKES Accountability Act” mandate the Department of Homeland Security to monitor and report on technologies containing Deepfake content, aiming to protect national security and provide legal remedies for victims affected by Deepfake technology.
Recent proposals include the “DEFIANCE Act of 2024” and the “Protecting Consumers from Deceptive AI Act”, designed to strengthen regulations and protections related to Deepfake technology.
Taiwan
In 2023, Taiwan passed amendments to Article 319-4 “Offense of False Image” and Article 339-4 of the Criminal Code, which include aggravated fraud for false images, sounds, or electromagnetic records. These amendments aim to curb digital gender-based violence or fraud crimes using Deepfake or other digital synthesis methods. The Institute for Information Industry has also recommended that the government, in addition to amending existing regulations to address emerging technology risks, should follow international trends by establishing principled guidelines for AI development. This includes creating AI governance guidelines for the government and businesses to follow, deepening AI governance.
Challenges in Regulating Deepfake Technology
Technical Challenges
One of the main challenges in regulating Deepfake technology is tracking and identifying the creators. Many Deepfake creators operate anonymously, making it difficult for authorities to trace the source of forged content, let alone enforce the law effectively. Moreover, as AI technology continues to advance, it may bypass existing safeguards, making it increasingly challenging for future investigations and regulatory efforts.
Balancing Free Speech and Regulation
Another major challenge is balancing the need for regulation with the protection of free speech. While it is crucial to curb the spread of misleading or harmful Deepfake content, overly strict regulation could infringe on free speech. This issue is particularly contentious in the context of political discourse, where Deepfake might be used for satire or personal expression.
Inadequacies of Existing Legal Frameworks
Due to the rapid evolution of AI technology, most existing laws were not designed with applications like Deepfake in mind, leading to gaps in enforcement or inconsistent regulations. As a result, international organizations like the World Intellectual Property Organization (WIPO) and the Electronic Frontier Foundation (EFF) emphasize the need for governments to develop comprehensive and flexible regulations specifically aimed at effectively managing Deepfake-related crimes.
Combating Deepfake Threats: Technology, Society, and Legal Synergy
1. Technological Solutions
Deploying advanced technological solutions is a crucial step in mitigating Deepfake-related risks. These solutions include detection mechanisms that use AI and machine learning to analyze inconsistencies in digital media or identify signs of manipulation to detect and flag Deepfake content, preventing harm from escalating. Digital watermarking and signatures further enhance content verification efforts by embedding unique identifiers in digital content to verify its originality and integrity. Additionally, with the rise in Deepfake misuse cases, real-time computer forensics software like Wireshark and EnCase can assist investigators and law enforcement in tracing the source of Deepfake content, providing crucial evidence for prosecution.
2. Public Awareness and Media Literacy
In addition to technical efforts, public awareness, and responsible media consumption are equally important in reducing Deepfake risks. Media literacy initiatives aim to educate the public on how to identify and evaluate digital content, seek reliable sources, and fact-check information to reduce their susceptibility to misinformation. In this regard, digital media platforms should take responsibility for fact-checking, flagging Deepfake content, encouraging users to choose verified content, and proactively reporting suspicious media to prevent the spread of Deepfake content.
3. Regulatory and Policy Initiatives
Finally, the implementation and enforcement of comprehensive regulatory policies and specific laws to curb misuse are the most critical components in preventing Deepfake threats. Regulatory measures should cover source verification, forgery, and dissemination of digital content, and it is recommended to enhance the adaptability of legal frameworks to categorize future uses and impose varying levels of obligations to flexibly respond to advances in AI technology.
For example, in April last year, the EU published a “Proposal for a Regulation on a European approach for Artificial Intelligence” (commonly referred to as the AI Act). Based on the level of risk and significance, the proposal classifies AI applications into four categories—“unacceptable risk, high risk, limited risk, minimal/no risk”—and requires corresponding obligations such as “prohibitions, information provision, usage recording, assisting regulatory oversight, AI action notifications, and warning labels.”
How Can Businesses Effectively Prevent Deepfake Threats?
Self-regulation in the fintech industry is a key factor in national efforts to combat Deepfake crimes. Businesses should actively adhere to technology and governance responsibility frameworks. For instance, last July, the White House announced voluntary commitments with seven tech giants—Amazon, Anthropic, Google, Inflection, Meta, Microsoft, and OpenAI—to develop AI systems based on principles of safety, security, and trust, ensuring a balance between technological advancement and societal safety.
Meanwhile, companies can adopt advanced AI verification solutions to enhance the security of digital services and increase user trust. Facing the growing risks of digital financial services, Authme tackles Deepfake threats from a fraud prevention perspective—using ISO 30107-certified Authme facial spoofing resistance technology, AI analyzes facial biometrics, including facial depth, skin texture, and microvascular flow, to determine if the person in front of the camera is real, helping businesses apply secure identity verification services to strengthen resilience and flexibility.
