Many professionals, including individual workers, unions, and organizations, are refusing to share their personal data—such as voices, texts, images, and faces—for training artificial intelligence systems. Their main concern is that by contributing their data, they are indirectly helping to develop tools that could replace them in their jobs. This pushback is largely about protecting their livelihoods and intellectual property. However, these efforts are caught in a complex web of ethical and structural challenges. Voice actors and dubbing professionals have been among the most vocal in resisting AI training. In France, 32 voice actors, including Pascale Chemin, who voices the character Wraith, refused to sign a contract addendum from Electronic Arts that would have required them to provide their recordings for AI training. They viewed this as a direct attempt to "train their own replacement." Similarly, in Germany, the Voice Dubbing Society (VDS) organized a mass boycott after Netflix tried to include clauses requiring dubbing actors to allow their voices to be used in AI models without extra pay. The actors called this a "legal act of self-erasure." Beyond voice actors, Hollywood actors and screenwriters have also resisted similar practices. Gal Gadot, for example, fought for six months to ensure that her contract for the film Bitcoin did not allow studios to use AI to alter or clone her performances. This stance was supported by the Screen Actors Guild–American Federation of Television and Radio Artists (SAG-AFTRA). In 2024–2026, thousands of screenwriters and voice actors went on strike, partly because they lacked protections against studios using their likenesses or texts to build AI substitution databases. Scientists and educators have also raised concerns about AI's role in academia. Over 3,200 French teachers, researchers, and mathematicians signed a manifesto opposing the use of generative AI in their work. They worry that AI could automate scientific research and diminish the value of human intellect. While environmental concerns are part of their reasoning, the broader issue is the push toward a future where AI replaces human creativity and reasoning. Despite these valid concerns, the refusal of individuals and groups to contribute data to AI training faces significant challenges. AI models are already trained on vast amounts of data from the internet, and removing the data of a few individuals or groups has little effect on the AI's ability to replicate styles or voices. Studies from institutions like MIT suggest that at a large enough scale, AI can still function effectively even without specific contributions. Moreover, by refusing to train AI, professionals risk allowing tech companies to use lower quality or pirated data to develop AI models. This could lead to the creation of less ethical, more impersonal AI that studios might use to replace human workers, potentially harming the industry as a whole. Additionally, many professionals who resist AI training still rely on AI-powered tools in their daily work, such as search engines or editing software. This creates a difficult ethical dilemma: where to draw the line between AI used as a tool and AI used as a replacement.