Freedom House Companies Companies should commit to respecting the rights of people who use their platforms or services and addressing any adverse impact that their products might have on human rights. The Global Network Initiative’s Principles provide concrete recommendations on how to do so. Companies should support the accessibility of circumvention technology and resist government orders to shut down internet connectivity or ban digital services. Service providers should use all available legal channels to challenge such requests from state agencies, whether they are official or informal, especially when they relate to the accounts of human rights defenders, civil society activists, journalists, or other at-risk individuals. If companies cannot resist demands in full, they should ensure that any restrictions or disruptions are as limited as possible in duration, geographic scope, and type of content affected. Companies should thoroughly document government demands internally, and notify people who use their platforms as to why connectivity or content may be restricted, especially in countries where government actions lack transparency. When faced with a choice between a ban of their services and complying with censorship orders, companies should bring strategic legal cases that challenge government overreach, in consultation or partnership with civil society. 2. DEFEND INFORMATION INTEGRITY IN THE AGE OF AI Even before the new wave of generative artificial intelligence (AI) products, AI was a key factor in the crisis of information integrity, serving as an intensifier in environments that were already vulnerable to manipulation. However, advancements in generative AI will supercharge the creation and dissemination of false and misleading content by state and nonstate actors, demanding a prompt response to safeguard access to reliable online information. Governments Governments should ensure that human rights principles, transparency, and independent oversight are embedded into AI regulation. Policymakers should specifically include robust protections against ineffective and unsafe systems, address algorithmic discrimination, require independent audits and human rights–based impact assessments, and mandate increased transparency regarding the design, testing, use, and effects of AI products. They should also require human review alternatives for AI decisions, such as in content moderation, and provide people with notice and clear explanations on how automated systems are being utilized. Governments should establish mechanisms for appeal and redress in cases of discrimination by AI systems. Finally, regulators should be empowered with sufficient resources and expertise to enforce their own rules and verify that companies are adhering to relevant laws. Specifically, the US government should follow through on an executive order—in development at the time of writing—that includes protections outlined in the Office of Science and Technology Policy’s Blueprint for an AI Bill of Rights, such as safeguards against algorithmic discrimination, limits on data use, and requirements for notice and explanation. In addition to such action, Congress should work with civil society and the executive branch to craft legislation that takes a rights-based approach to AI governance and transforms guiding principles into binding law. The US Federal Election Commission should also prohibit political parties, committees, and candidates from intentionally misrepresenting candidates in advertising that features AI-generated or manipulated imagery. In Europe, lawmakers negotiating on the proposed EU AI Act should, at minimum, ensure that the final text obligates companies to label AI-generated media and conduct fundamental rights impact assessments for uses of AI services that present risks for human rights. @freedomhouse freedomhouse.org 31