Freedom House AI REGULATION IN PRACTICE Much of the policy conversation about the AI boom has centered around the value of comprehensive regulatory frameworks and safety concerns. Recent regulation has taken a narrower approach aimed at solving particular issues, such as impersonation or harassment. Some measures raised concerns about overreach, such as in India and Mexico, as described below. Broader concerns persist about how AI models can reinforce censorship practices of authoritarian regimes like those that rule Russia and China. Child protection: Lawmakers sought to limit children’s access to AI services deemed especially sensitive. For example, Australia’s eSafety commissioner announced in September 2025 that the country’s under-16 age-assurance framework would apply to certain AI companies, particularly those that host “AI companions” designed to simulate friendship or romantic relationships. While the privacy and security risks around age assurance persist, early research on AI companion services indicate they threaten emotional and psychological harms for vulnerable users, particularly minors. Online harassment: Bad actors abuse AI tools to generate sexual imagery of people without their consent. Governments have rightly taken up efforts to push tech platforms to remove these images and create barriers to their generation. UK authorities amended a law penalizing people who create or commission nonconsensual imagery to include AI tools in February 2026, and launched an investigation into X’s Grok service after the platform failed to remove such images, including abuse material depicting children. Labeling: In February 2026, the Indian government passed new rules requiring technology companies and social media platforms to ensure that “synthetically generated information” content features prominent labels. While labeling AI-generated content is a positive step, the new law also mandated social media platforms to remove any content flagged by the authorities within 3 hours, reduced from 36 hours under previous law. The Indian government has built expansive infrastructure for internet censorship, prompting concerns from civil society that the new rules would be abused to silence online content deemed unfavorable by the government, including independent journalism and satire, particularly because the 3-hour period limits careful evaluation of reported material, forcing platforms to err on the side of removal. Impersonation: Policymakers around the world expressed concern about how generative AI tools could be used to impersonate people to cause reputational harm or carry out scams. In some countries, lawmakers sought to criminalize such impersonation. For example, in Mexico, the state of San Luis Potosí reformed the penal code to bar unauthorized AI-generated use of a person’s image or voice, as well as the intentional dissemination of disinformation generated through AI. Despite a carveout for journalism, authorities used the law to arrest and detain two online journalists in May 2026, apparently because they reported on a video in which a local gang accused the state’s governor of corruption; the governor said the video was AI-generated. freedomhouse.org 11