# Can Laws Stop Deepfakes? South Korea's June 3 Elections Will Provide the First Real Answer


South Korea is about to conduct an unprecedented experiment in deepfake regulation. On June 3, the nation will hold local elections under two new laws specifically designed to combat AI-generated synthetic media in political campaigns—marking the first comprehensive legal stress test of whether regulations can actually contain the threat that generative AI has unleashed on democratic processes.


The question at stake goes far beyond South Korea's local contests: can any legal framework effectively police deepfakes when the technology is trivially accessible, distributed through encrypted channels, and evolving faster than regulators can respond?


## The Threat: Deepfakes Go Mainstream


Deepfakes are no longer the province of sophisticated threat actors. What once required extensive technical expertise and specialized hardware can now be generated by anyone with a smartphone and a free consumer application in a matter of minutes.


The evidence is mounting globally:


United States (2024): New Hampshire voters received a robocall impersonating then-President Joe Biden, urging residents not to vote in the state primary. The AI-generated audio was convincing enough to confuse recipients and sow doubt about election integrity.


South Korea (2022): Ahead of the presidential election, a widely-circulated video falsely depicted opposition leader Lee Jae-myung ending his hunger strike—at a moment when he was both the frontrunner and head of his party. The video spread rapidly across social media before being debunked.


South Korea (Multiple incidents): Deepfake videos of political candidates have circulated on social media. Most disturbing: AI-generated television news reports fabricating statements from public figures have been distributed, blurring the line between misinformation and synthetic media.


Brian Long, CEO and co-founder of Adaptive Security, a firm specializing in threat awareness training for phishing and deepfakes, provided critical context to Dark Reading: "Two years ago, generating a convincing deepfake required real technical skill. Today, consumer tools produce convincing audio, video, and synthetic text in minutes."


That democratization of capability is the core problem regulators face.


## The Technical Reality: Why It's So Easy Now


The explosion in accessible deepfake creation stems directly from advances in generative AI:


  • Generative Adversarial Networks (GANs) and diffusion models now produce video and audio so convincing that human judgment alone cannot reliably detect them
  • Text-to-speech systems can mimic individual voices with just 30 seconds of source audio
  • Face-swapping tools run on commodity GPUs and are freely available on GitHub
  • Pre-trained models eliminate the need to build systems from scratch

  • The barrier to entry has collapsed. A moderately technical person can generate a convincing deepfake in under an hour. A savvy campaign operative could do it in minutes.


    ## South Korea's Regulatory Response: The Two New Laws


    South Korea's government has enacted two laws to address this threat ahead of the June 3 elections:


    1. Election Law Amendment: Criminalizes the creation and distribution of deepfakes intended to influence elections

    2. Specific Framework: Imposes obligations on social media platforms and messaging services to detect, report, and remove violating content


    ### Key Provisions:


    | Element | Description |

    |---------|-------------|

    | Scope | Applies to synthetic media created with intent to influence voting behavior |

    | Penalties | Criminal liability for creators; potential platform liability for slow removal |

    | Detection Requirement | Platforms must implement monitoring and reporting mechanisms |

    | Speed Requirement | Content must be removed within a defined timeframe (typically 24 hours for election-related material) |

    | Enforcement | Election authorities and law enforcement agencies coordinate investigations |


    According to Long, the value of clear legal infrastructure is not just punitive—it's procedural. "Clear legal infrastructure allows investigators to act faster while giving platforms clearer obligations to remove violating content," he explained. When platforms know exactly what they're responsible for, and when investigators have explicit legal authority, responses accelerate.


    ## The Critical Limitations: Where Regulation Breaks Down


    However, the regulatory framework faces a serious blind spot: encrypted channels and private messaging platforms.


    Long identified this gap directly: "The problem exists in the channels that regulators can't clearly reach. Deepfakes distributed through encrypted messaging apps, targeted SMS campaigns, and closed group chats operate in a regulatory gray zone."


    This is not a minor loophole. Political operatives have long understood that the most effective disinformation spreads through private channels—WhatsApp groups, Telegram, encrypted Signal chats, and SMS text blasts. These channels are harder to monitor, harder to trace, and much harder to regulate without access to private communications.


    ### Additional Challenges:


  • Speed vs. Detection: Even with legal obligations, platforms cannot catch all deepfakes before they spread. A video can go viral in hours; detection and removal may take days.
  • Definition Ambiguity: What constitutes a "deepfake" versus satire, parody, or edited content? The law's language matters enormously, and litigation will clarify boundaries.
  • Enforcement Across Borders: If a deepfake is created outside South Korea, does the law apply? Who enforces it?
  • False Positives and Censorship Risk: Overly aggressive removal could suppress legitimate political speech and satirical content.

  • ## The Broader Implications: Can Any Nation Regulate Deepfakes?


    South Korea's June 3 election serves as a test case for every democracy grappling with synthetic media. The results will inform policy in the United States, European Union, India, and beyond.


    If South Korea's laws prove effective, they may become a template—demonstrating that legal clarity, platform accountability, and swift enforcement can suppress deepfake distribution during high-stakes elections.


    If they fail—if deepfakes circulate widely despite legal prohibitions—then policymakers will face a harder question: whether regulation is sufficient, or whether the problem requires technological solutions (detection algorithms, blockchain verification, digital signatures on authentic content).


    ---


    ## HackWire Analysis


    South Korea's June 3 election represents a critical inflection point not just for deepfake policy, but for the broader question of whether law can outpace technology. The regulatory framework is sound in principle—it criminalizes harmful conduct and assigns clear responsibility to platforms. But the architecture of modern communication defeats it at a fundamental level.


    The unspoken reality is this: encrypted messaging is a regulatory dead zone, and that's where the most damaging political deepfakes will live. A campaign operative can create a convincing video of a rival candidate in 15 minutes, distribute it to 50,000 voters via WhatsApp in another 15 minutes, and delete evidence of creation in another 5. The victim has hours to debunk it, and regulators have no mechanism to prevent it.


    South Korea's laws are aggressive and well-intentioned. But they optimize for detection on public platforms—Twitter/X, Facebook, YouTube, public Telegram channels—where scale and visibility make moderation more feasible. They do nothing about the channels where political operatives actually work: encrypted, private, ephemeral.


    The real question is not whether these two laws will work, but whether South Korea will discover that deepfake regulation requires treating encrypted messaging differently than it currently does. That's a thornier policy problem than South Korea's lawmakers acknowledged. For every democracy that watches this election carefully, that's the hard lesson coming.


    The June 3 election won't be a failure of South Korea's legal framework. It will be a failure of any single nation to regulate communications that cross borders instantly and operate in encrypted channels beyond its reach. That's not a regulatory problem. It's an architectural problem. And it won't be solved by law alone.


    HackWire Editorial


    ---


    ## Related Coverage


  • Read more in our [Vulnerabilities](https://www.hackwire.news/category/vulnerabilities) coverage
  • Cross-reference with [Breaches](https://www.hackwire.news/category/breaches) and [Malware](https://www.hackwire.news/category/malware)
  • Stay current via the [HackWire homepage](https://www.hackwire.news/)