가상 인플루언서 산업이 급성장하며 브랜드 마케팅의 핵심으로 자리 잡고 있으나, EU 인공지능 법안 등 글로벌 규제가 강화되면서 실무자들이 큰 혼란을 겪고 있습니다. AI 생성 콘텐츠 표시 의무가 확대되고 있지만, 법의 모호성으로 인해 제작자와 브랜드 담당자들은 대응 방안을 찾지 못해 위기를 맞이하고 있습니다.
번역된 본문
클라리사 만스브리지(Clarissa Mansbridge)는 제이슨 데룰로(Jason DeRulo)나 킴 카다시안(Kim Kardashian)의 전 경호원인 셩고 딘(Shengo Deane) 같은 인물들과 함께 일하며 약 20년 동안 무대 뒤에서 엔터테인먼트 경력을 쌓았지만, 공개 행사에 모습을 드러내는 일은 거의 없었다. 그녀는 누군가의 브랜드를 구축하고, 위기를 피하며, 일상적인 문제들을 해결하는 짜릿함을 즐겼다.
그러던 3년 전, 그녀의 삶은 송두리째 뒤바뀌었다. 임신 33주 차에 그녀의 아들이 심장이 뛰지 않는 상태로 태어난 것이다. 의사들의 응급조치로 아들을 살려냈지만, 중환자실에서 보낸 몇 달 동안은 아들이 살아남을 수 있을지 알 수 없는 끔찍한 순간들의 연속이었다. 기적적으로 집으로 돌아올 수 있었을 때, 호주 브리즈번에 거주하는 만스브리지는 아들의 회복과 관련된 합병증으로 인해 자신의 근무 환경을 바꿔야 한다는 것을 깨달았다.
그녀는 AI 이미지 생성 도구를 실험하기 시작했고, "겸손하게 부유하며 건전한" 인플루언서이자 모델인 '미아 메타버스(Mia Metaverse)'를 탄생시켰다. 초기 심즈(Sims) 게임의 팬이었던 그녀는 미아가 활보할 매력적인 배경과 스토리라인을 만드는 일에 흠뻑 빠졌다. 34세인 만스브리지는 현재 브랜드 협업과 기업들을 위한 맞춤형 AI 인플루언서 디자인을 통해 월 약 6,000달러(한화 약 820만 원)를 벌고 있다고 말한다. 이들 중 다수는 비밀유지계약(NDA)을 요구하므로 그녀가 어떤 브랜드와 일했는지 공개할 수 없다.
하지만 론칭 이래 가장 수익이 높은 해임에도 불구하고, 만스브리지와 다른 AI 크리에이터들은 소셜 플랫폼에서 AI 콘텐츠가 다뤄지는 방식의 글로벌 변화가 그들의 비즈니스 모델을 뒤흔들 수 있다고 우려하고 있다. 2022년 생성형 AI(generative AI)와 텍스트-투-이미지(text-to-image) 도구가 등장한 이래, 크리에이터들은 더 저렴하고 확장 가능하며 브랜드에 안전한 가상 인물을 중심으로 수익성 높은 계정을 구축해 왔다.
약 145억 달러 규모로 추산되는 가상 인플루언서 시장은 연간 33.6%의 성장률을 기록하며 2033년까지 1,104억 달러 규모에 도달할 것으로 전망된다. 브랜드들 역시 확실하게 참여하고 있다. 소셜 에이전시인 '빌리언 달러 보이(Billion Dollar Boy)'의 연구에 따르면, 79%의 마케터가 AI가 생성한 크리에이터 콘텐츠에 대한 투자를 늘리고 있다.
하지만 이 새롭게 싹트고 있는 크리에이터 산업에 규제가 집중되면서 흐름이 반전될 수도 있다. 8월 2일부터 시행된 'EU 인공지능 법안(EU AI Act)'의 새로운 규정에 따라 AI가 생성하거나 조작된 홍보 콘텐츠는 명확하게 표시되어야 한다. 또한 틱톡(TikTok)과 같은 플랫폼들 역시 감지 시스템을 강화하고, 합성 스팸보다는 인간 크리에이터의 게시물을 우선시하도록 피드를 개편하고 있다. AI 인플루언서가 생계인 사람들에게 이는 헤쳐 나갈 명확한 지침이 없는 규제의 미로와 같은 위험을 의미한다.
EU AI 법안은 역사상 최초로 포괄적인 AI 법적 체계로, 유럽연합 집행위원장 우르줄라 폰데어라이엔(Ursula von der Leyen)이 표현한 '세계 최초'의 기준이며 유럽에 '신뢰할 수 있는 AI'를 조성하는 것을 목표로 한다. 이 법안 제50조에 따라, AI 시스템의 제공자 및 운영자는 오디오, 이미지, 비디오 또는 텍스트 콘텐츠가 인공적으로 생성되거나 조작되었을 때 이를 명시해야 한다.
단, "해당 콘텐츠가 예술, 창작, 풍자, 허구 또는 유사한 작업의 일부로 명백히 구성된 경우" 투명성 의무는 작품의 전시나 감상을 방해하지 않는 수준의 공개로 제한된다. 이는 수많은 해석의 여지와 혼란을 남긴다. 이 법안의 협상자이자 의회 의원인 악셀 보스(Axel Voss)의 디지털 정책 보좌관인 카이 제너(Kai Zenner)는 이 법안이 너무 이르고 너무 추상적으로 작성되었다고 말한다.
그는 "ChatGPT가 막 등장했을 때였으며, 국회의원들이 우리가 규제하려는 대상에 대해 충분한 지식이 없었다"고 설명한다. "그저 AI 시대가 도래했고 더 많은 AI 생성 콘텐츠가 나올 것이니, (라벨링 조항을 시행하여) EU 시민들의 AI 리터러시(이해력)를 높여보자는 것이 추진력이었습니다."
더 많은 법률 제정이 이러한 불확실성을 가중시킬 전망이다. 제너는 "올해 하반기와 내년에 여러 차례의 개정이 이루어질 예정이며, 이는 인플루언서들에게 콘텐츠 표시를 의무화할 뿐만 아니라, 콘텐츠의 출처, 목적, 재원 조달 방식에 대한 투명성을 제공하도록 강제할 것"이라고 설명한다. "아직도 그것이 어떻게 관리될 것인지는 정말 불분명합니다."
Comment Loader Save Story Save this story Comment Loader Save Story Save this story Clarissa Mansbridge spent nearly two decades building an entertainment career behind the scenes, working with the likes of Jason DeRulo and Shengo Deane, Kim Kardashian’s former bodyguard, but she rarely appeared at public events. She liked the thrill of it: building someone’s brand, swerving crises, and dealing with day-to-day challenges. Then, three years ago, her life was upended when her son was born without a heartbeat at 33 weeks. Doctors managed to resuscitate him, but the months in intensive care brought scare after scare, with no certainty that he would survive. When, eventually, miraculously, they made it home, Mansbridge, who lives in Brisbane, Australia, knew her work life would have to change due to complications related to her son’s recovery. She began experimenting with AI image-generation tools and built the “humbly wealthy, wholesome” influencer and model Mia Metaverse . A fan of early Sims, she reveled in crafting storylines and glamorous settings for Mia to move through. Mansbridge, 34, says she now earns around $6,000 a month through brand collaborations and designing bespoke AI influencers for brands, many of which require her to sign an NDA so she can’t disclose which ones she’s worked for. Yet despite this being her most profitable year since launching, Mansbridge and other AI creators worry that global changes to how AI content is handled on social platforms could upend their business model. Since the emergence of generative AI and text-to-image tools in 2022, creators have built lucrative accounts around virtual personalities that are more affordable, scalable, and brand-safe. Estimated to be worth $14.5 billion , the virtual influencer market is projected to reach $110.4 billion by 2033, with a reported 33.6 percent annual growth rate. Brands are firmly on board: Research by the social agency Billion Dollar Boy found that 79 percent of marketers are increasing their investment in AI-generated creator content. But as regulation converges on this burgeoning creator industry, the tide may be turning. From August 2, new rules under the EU Artificial Intelligence Act require AI-generated or -manipulated promotional content to be clearly labeled, while platforms like TikTok are tightening detection systems and reshaping feeds to elevate human creators over synthetic spam. For people whose livelihoods depend on AI influence, the risk is a regulatory maze, with no clear road map to navigate it. The EU AI Act is the first-ever comprehensive legal framework on AI—a “ global first ,” as European Commission president Ursula von der Leyen called it—and aims to foster “trustworthy AI” in Europe. Under Article 50 of the act, providers and deployers of AI systems must disclose when audio, image, video, or text content has been artificially generated or manipulated. But “where the content forms part of an evidently artistic, creative, satirical, fictional, or analogous work,” the transparency obligations are limited to disclosure that doesn’t hamper the display or enjoyment of the work. That leaves plenty of room for interpretation—and confusion. Kai Zenner , digital policy adviser to Axel Voss, a member of Parliament and a negotiator on the Act, says it was drafted too early, too abstractly. “ChatGPT had only just come out, and there was scant knowledge among MEPs and parliamentarians about what we were regulating for,” he explains. “The driver was simply: AI is here and more AI-generated content is coming, so let’s raise the AI literacy of EU citizens [by implementing the labeling clause].” More legislation is likely to add to the uncertainty. “There are several revisions coming later this year and next that will force influencers not only to label their content, but provide transparency on where they come from, what they want to do, and how they are financed,” explains Zenner. “It’s still really unclear how that will be managed, since all we have to go on is Article 50.” Zenner worries that AI creators could be pulled into a tangle of overlapping rules: the AI Act, copyright law, audiovisual media rules, consumer protection law, and possibly even the Digital Fairness Act, the EU's proposed update to consumer protection rules for the digital age, which targets online tactics that exploit or mislead users. “People running these accounts could find one law says their content is lawful, while another law says it's unlawful,” he says. Swinging At AI Content With A Sledgehammer With 60 percent of creators admitting in 2025 that their AI-assisted content had broken industry rules because of mislabeling, the risk of missteps is only set to grow in 2026. Many major social media platforms have introduced crackdowns of their own. In mid-July, TikTok announced a detection system to catch AI spam, which it claims is “crowding out original creators.” It already requires creators to label AI-generated content and lets users reduce the amount of AI content in their feeds. Pinterest takes a similar approach, while Meta now adds an “AI info” label across Facebook, Instagram, and Threads when it detects AI-generated content. “The same platforms that produced more content, faster, and at a lower cost, are now tweaking their algorithms to push that content back down, but none of these systems are precise enough,” explains Matt Navarra, social media consultant and author of the weekly Geekout Newsletter . “The platforms are swinging at AI content with a sledgehammer when what they need is a scalpel.” He believes human creators will get caught out, even if they used AI for minor reasons, like fixing up their grammar. Others are already seeing the cost of being too visible to algorithms. Lionel Metge, a hip-hop artist and producer who created a reggaeton/Afrobeat AI artist named Zairo at the start of this year, has already felt the shift in how platforms handle synthetic content. He’s noticed that the new detection systems can identify AI-generated content as soon as it’s published, especially when an account labels itself as a digital creation. Metge, based in Lyon, accepts that the rules could already be alienating Zairo’s 23,800-follower account and some of its target audience, but he believes they’re necessary to avoid misleading the public, especially as generation models become ever more realistic. “I hope in future that algorithms won’t overly restrict the reach of AI-generated content,” he tells WIRED, adding that he plans to double down on “quality and overall coherence” to keep Zairo standing out. It’s a moot point for experts operating in the creator industry. Fanvue , a platform where creators connect with fans and monetize their content, has been working with creators on disclosure standards ahead of the August deadline. “We’re seeing creators lean into transparency as a differentiator, with some building unique content that’s only possible for an AI persona, rather than trying to conceal it,” says Rob Haywood, Fanvue’s head of safety. He believes in-platform regulation is a response to a real problem: TikTok is responding to “a tide of low-effort synthetic content flooding feeds and reducing the visibility of real creatives.” Zenner advises creators to get ahead of future legislation by showing “their best efforts,” which could help if enforcement becomes tougher or less clear. Mansbridge is building beyond a single character, towards a portfolio of AI personas that she hopes can be cross-promoted and eventually, run autonomously. “I want to create a legacy for my son, so that if ever he needs anything, it’s financially covered,” she explains. Her approach isn’t to hide the AI, but to make it safer to sell. She built Mia Metaverse to be “relatable, but aspirational,” a clean slate with no scandals, politics, or morals clauses. In a market facing tougher transparency rules, that kind of predictability could become an asset: The more platforms and regulato