Recently, the voice of Taiwan President Lai Ching-te was allegedly used in an AI-generated video, sparking significant controversy. The incident has triggered a debate about freedom of speech vs criminal offenses linked to deepfake technology.
Critics like Taiwanese Legislator Fan Yun who’s the ruling Democratic Progressive Party (DPP) caucus secretary-general said that freedom of speech must absolutely be protected, but that it does not include fabricated defamation, the Central News Agency reported.
Senior legal experts stated that deepfake technology is not inherently illegal. However, if someone impersonates a specific individual’s voice, appearance, or statements in a way that causes the public to mistakenly believe the content was created or delivered by that person, it could potentially lead to criminal liability for offenses such as forgery, fraud, and defamation, or constitute violations of personality rights, image rights, and personal data protection regulations.
Legal experts: AI impersonation of another person’s voice could lead to criminal liability
Taiwan’s Criminal Investigation Bureau (CIB) said on July 26 that it had received public complaints about a video titled “Oil Cannot Defeat You” (油不得你), which allegedly used AI-generated audio to imitate Lai Ching-te’s voice and potentially mislead viewers into believing the president had made the statements himself, the Taipei Times reported. Authorities have opened an investigation into the case.
The video, titled “Oil Cannot Defeat You,” was created and recently released by “Principal Lai’er” (萊爾校長) creator Wei Chun-you, a former director of the KMT New Media Department. Wei received support from the KMT camp after being questioned by police over the production of the video.
For the best of our weekly content！
You are now signed up for our newsletter
Check your email to complete sign up
According to the Central News Agency (CNA), a senior legal expert with 20 years of professional practice experience said that Taiwan’s Legislative Yuan’s Internal Administration Committee held a special hearing in 2021 on risks arising from AI deepfake technology, including AI-related fraud, non-consensual sexual private images, election-related misinformation and fabricated evidence.
Experts noted that, under Article 220 of Taiwan’s Criminal Code, audio recordings, video recordings, or electronic records that possess the function of expressing information or serving as proof may legally be treated as “documents”under the law.
If AI is used to fabricate or alter the audio or video of a specific individual, causing others to mistakenly believe that the person personally made the statements or acknowledged certain content, and if such actions are sufficient to cause harm to the public or another person, the act could potentially involve Article 210 of the Criminal Code — forgery or alteration of private documents.
If the fabricated audio or video is then uploaded online, provided to media outlets, or used publicly, it could also potentially involve Article 216 — the use of forged documents.
Experts emphasized that not all AI-generated audio or video content would automatically be considered forged documents. The assessment must consider factors including whether the audio or video has a proof-related function; whether it impersonates another person’s identity; whether the content has been fabricated or altered; and whether it may cause actual harm.
If deepfake technology is used for purposes such as financial fraud, creating false imagery, or manipulating elections, it may involve additional legal liabilities, including aggravated fraud, violations of sexual privacy rights, defamation, and breaches of election and recall laws.
Lawyer Lan Tian: AI applications cannot ignore authorization issues
Attorney Huang Xiulan, known as “Lawyer Lan Tian,” pointed out that even when AI-generated voices are created for artistic expression or creative performances, voices, voiceprints, and images still involve personality rights, portrait rights, and personal data protection issues.
Huang told CNA, “AI applications cannot ignore authorization issues.”
Therefore, before using a real person’s voice to create an AI model, prior authorization from that individual must be obtained. Even in cases involving the voices or images of deceased artists, their wishes during life and the positions of their family members should be respected. Creative purposes, parody, or discussion of public issues cannot be used as a reason to disregard the legitimate rights and interests of the individuals concerned.
Freedom of speech does not include fabricated defamation
Fan Yun emphasized freedom of speech within legal framework at a party caucus press conference, according to CNA.
“Freedom of speech must absolutely be protected, but it absolutely does not include fabrication or defamation,” she said, adding that whenever someone files a complaint, the judicial system will take corresponding action.
Fan said that international research has found that fabricated content and misinformation can harm a core element of democratic societies i.e. “trust.”
Fan Yun pointed out that “this issue is not targeting any particular political party or political figure; ordinary citizens can also be harmed by fabricated content and misinformation.” She urged opposition parties to look beyond partisan interests when addressing the issue. She added that evidence should be collected in accordance with Articles 210, 216, and 220 of the Criminal Code (crimes involving the forgery of documents and seals), and that whether this specific case constitutes a violation of the law should be determined through judicial proceedings.
DPP spokesperson Wu Zheng said at a DPP press conference on July 27 that using AI-generated content or deepfake technology to impersonate another person’s images, voice, or documents could potentially involve the crime of forging private documents, the Taipei Times reported.
He said that disputes caused by impersonation targeting public figures or political figures impose significant social costs. Therefore, whether in cases of fraud or misinformation-based cognitive warfare, the public needs to remain vigilant.
DPP legislator Chung Chia-pin said, “In the past, everyone agreed that deepfake videos were unacceptable. Whether it is a head of state or an ordinary citizen, everyone’s portrait rights and personality rights should be protected.”
He argued that if society now tolerates behavior by “Principal Lai’er” that goes beyond the scope of freedom of speech, then in the future, when ordinary citizens are similarly altered, fabricated, or targeted through dissemination of manipulated content, they may have nowhere to turn for help. He urged opposition parties not to set the wrong precedent.
False information could interfere with elections
DPP legislator Chen Pei-yu said on Facebook on July 26 that every content creator using AI or digital technologies should understand the relevant legal responsibilities.
She said that if someone uses AI technology to deliberately create content with a highly similar voice or image without obtaining the consent or authorization of the person concerned, and causes viewers to believe that the content reflects a genuine statement made by that person, it could potentially involve legal issues such as document forgery.
Chen pointed out that forgery of documents is a non-complaint-based offense (a public prosecution offense), meaning the victim does not necessarily need to personally file a complaint. Once police or prosecutors receive reports and obtain relevant evidence, they may investigate the matter in accordance with the law.
She said, “Whether it is the president, the heads of the five branches of government, administrative officials or ordinary people, you do not have to be a public figure. When such content affects your social credibility and undermines the public’s confidence in you, the state must intervene. This not only maintains social stability but also prevents people from having to bear responsibility for words or statements they never made.”
Chen added that police asking the creator of the reported video to provide an explanation is a lawful process to clarify facts and should not automatically be interpreted as suppression of freedom of speech.
Chen warned that if large quantities of low-cost AI-generated fake videos, images, and audio recordings, combined with overseas misinformation networks, spread across social media platforms before the year-end local elections, citizens may find it increasingly difficult to distinguish truth from falsehood. She cautioned that anyone could become a target of smears and attacks, potentially causing voter apathy and reducing willingness to participate in elections.
In response to criticism from KMT legislator Hsu Chiao-hsin regarding police handling of the case, Chen Pei-yu argued that such remarks were an attempt to shift attention away from the issue of AI impersonation and legal responsibility. She said that invoking phrases such as “checking water meters” (查水表)—a term carrying historical associations with political intimidation and official harassment—was being used to frame the issue politically as “the DPP suppressing freedom of speech,” thereby continuing to accumulate hostility and social division.
Political commentator Grace Wu questioned, “Is it acceptable to use a fabricated voice to commit fraud? Does the Blue Camp now support scams involving altered voices?”
Original article
Comments