The Free Flow — March 20, 2025
Meta tests crowd-sourced fact-checking, Texas scales back AI regulations, UNESCO critiques Nepal’s social media bill, China mandates AI content labeling, and more.
The Digital Age
» Meta to Test ‘Community Notes’ Fact Checking Tool:
Meta announced it would begin testing its crowd-sourced fact-checking program, Community Notes, on March 18.
Background:
In January, Meta discontinued its traditional fact-checking program over concerns of political bias
The community notes system allows users to add contextual notes to posts, aiming to provide additional information while avoiding censorship and top-down control.
Read more about community notes and Meta’s policy changes in Jacob Mchangama’s analysis at Persuasion.
» Texas Scales Back Proposed AI Regulation Bill:
Texas Responsible Artificial Intelligence Governance Act (TRAIGA) bill H.B. 1709 has been replaced with H.B. 149 (with the same name), stripping away many of the burdensome regulations from the original act and instead focusing on restricting government use of AI.
The bill prohibits state agencies from using AI to implement “social scoring” or to make decisions based on residents’ “social behavior” or “personal characteristics.”
A new provision forbids AI from facilitating political viewpoint discrimination online, which mirrors Texas’ social media law that was struck down as unconstitutional.
Although the bill avoids heavy-handed regulation on businesses and AI development, a lawsuit from industry groups is anticipated if the legislation passes in its current form.
» UNESCO Criticizes Nepal’s Social Media Bill:
After organizing a meeting between members of the Nepali Parliament and civil society, UNESCO has published an official review of Nepal’s Social Media Bill 2081, written by The Future of Free Speech’s Senior Legal Fellow Joan Barata.
Key Issues:
UNESCO’s review expressed concern about overlapping definitions of “platform” and “social media” in the bill, which could expand its reach beyond digital applications to encompass internet infrastructure as well.
Moreover, the bill mandates social media platforms obtain licenses, with conditions that may impose prior restraint
The bill also assigns regulatory oversight to a government body without ensuring its independence, violating international judicial standards.
Joan’s analysis recommended Nepal clarify definitions, ensure independent regulation, and reassess licensing and content restrictions.
For a detailed look into free speech concerns with Nepal’s law, you can check out the February 13 edition of The Free Flow.
» NewsGuard Launches ‘FAILSafe’ to Combat AI Disinformation:
NewsGuard has introduced the Foreign Adversary Infection of LLMs Safety Service (FAILSafe) to protect large language models (LLMs) from foreign influence operations.
Key Components:
Disinformation Narrative Feed: Provides real-time updates on false narratives propagated by foreign influence operations.
Influence Domain Dataset: Maintains a database of websites and accounts disseminating state-sponsored information.
Red-Teaming Services: Offers stress-testing of AI models to assess their susceptibility to foreign disinformation.
» OpenAI Calls for Ban on China’s DeepSeek AI:
Open AI has urged the U.S. to prohibit the use of AI models developed by DeepSeek, a Chinese AI lab, labeling it as “state-subsidized” and “state-controlled”
Concerns over state influence, intellectual property misuse, and data privacy issues underscore Open AI’s comments.
» China Mandates Labeling of AI-Generated Content:
On March 14, 2025, four Chinese government departments jointly issued a notice requiring explicit labeling of AI-generated content. This applies to AI-generated text, images, audio, video, and virtual scenes.
Requirements:
AI-generated content must include clear indicators, such as text, sound, or graphics, to inform users of its synthetic nature
Embedded metadata should contain information about the content’s AI-generated status, the service provider’s identity, and a unique content identifier
» Google Responds to Trump’s “AI Action Plan”:
In response to the Trump administration’s call for a national “AI Action Plan,” Google has published a policy proposal recommending weakened copyright restrictions and balanced export controls to promote AI.
Background:
Trump has been fiercely critical of the Biden administration’s approach to AI regulation, having revoked an executive order that required tech companies to brief the government about their AI technology before releasing it to the public.
Google’s policy proposal came shortly after OpenAI’s published recommendations—which include banning foreign AI platforms, such as China’s DeepSeek, on grounds of foreign censorship.
Google’s manifesto asks for more government-funded R&D—even as Elon Musk’s DOGE cuts back on federal research grants–and looser usage liability obligations, which force developers to bear some liability for how consumers use their AI models.
They also requested copyright law exemptions for AI models—as Google already faces heat for its Gemini chatbot’s ability to remove watermarks from images—and due to ongoing concerns that AI models are trained using copyrighted material.
It remains to be seen how much Trump’s AI Action Plan will incorporate Google and OpenAI’s recommendations as the UK and the EU pursue their own “consultations” with tech firms on revamping their AI policies to ensure global competitiveness.
» House GOP Leaders Subpoena Big Tech for AI “Censorship”:
The Republican Chairman of the House Judiciary Committee, Jim Jordan, subpoenaed 16 tech firms—including Alphabet, Meta, Microsoft, and OpenAI—to investigate allegations of “censorship” against conservatives in the moderation and output of AI models.
Background:
Just last week, Jim Jordan asked for Google and Facebook parent company Alphabet’s communications with the Biden administration while earlier requesting emails between eight social media firms and foreign governments.
The subpoenas follow longtime conservative complaints about censorship on social media platforms.
However, these new subpoenas target a wider range of tech companies than before—requesting emails from software firms like Palantir and chip-designer NVIDIA—and focus on AI training and content moderation specifically.
Ahead of the subpoenas, representative Jordan referenced a Judiciary Committee report from December 2024, which had found Biden officials “pressuring private companies to ‘advance equity,’ stop ‘algorithmic discrimination,’ and ‘mitigate the production of harmful and biased outputs.’”
The subpoenas ask for all emails between tech companies and government officials between January 2020 and January 2025 “referring or relating to the moderation, deletion, suppression, restriction, or reduced circulation of the content, input, or output of an AI model, training dataset, algorithm, system, or product.”
The Brussels Effect: Europe and Beyond
» French Publishers Sue Meta:
The National Publishing Union, National Union of Authors and Composers, and Société des Gens de Lettres sued Meta for copyright infringement in France.
Background:
Meta has deployed generative AI chatbots across Facebook, Instagram, and WhatsApp.
The French trade groups argue Meta’s chatbots are trained using members’ copyright-protected material.
Moreover, they allege Meta’s chatbots produce “fake books” by recycling published, copyrighted content.
The French authors and publishers’ case is the latest in a series of recent media clashes with AI companies regarding copyright law.
In the US, Thomson Reuter’s successful litigation restricted AI’s ability to use copyrighted material under fair use, while in the UK more than 1,000 artists released a silent album last month opposing the country’s potential loosening of copyright regulations for AI models.
» Spain Proposes Heavy Fines for Unlabeled AI-Generated Content:
Spain’s government has approved a draft law imposing substantial fines on companies that fail to label AI-generated content appropriately.
The bill:
Non-compliance would be classified as a “serious offense,” with penalties reaching up to €35 million or 7% of a company’s global annual turnover.
The legislation would ban the use of imperceptible sounds or images to manipulate groups and restrict organizations from using AI to classify individuals based on biometric data to assess risks or grant benefits.
The legislation aligns with the EU’s AI Act, employing strict transparency obligations for high-risk AI systems.
» UK Promises Huge Fines for Social Media Companies that Fail to Comply with Online Safety Act:
The UK’s Online Safety Act has come in into force, mandating online platforms proactively prevent and swiftly remove illegal content.
Background:
The Office of Communications is responsible for overseeing compliance.
Affected platforms, including messaging apps, file-sharing services, and social media networks, are required to implement robust moderation systems, facilitate user reporting mechanisms, and conduct thorough risk assessments to mitigate the presence of illegal content.
Failing to adhere to requirements could result in fines of up to £18 million or 10% of their global revenue, whichever is higher.
Press Freedom Watch:
President Trump Accuses Media of 'Illegal' Reporting: In a recent address at the Department of Justice, President Donald Trump accused certain media outlets, such as CNN and MSNBC, of engaging in "illegal" reporting, labeling them as "political arms of the Democratic Party." He suggested that press criticism of judges should be criminalized, despite his own and his allies' frequent attacks on the judiciary.
Peru’s Amendment to APCI Law Targets Civil Society: On March 12, 2025, Peru’s Congress approved amendments to the Law of the Peruvian Agency for International Cooperation (APCI). NGOs are now required to obtain approval from APCI for their plans, programs, and activities, and the law imposes fines up to $720,0000 for NGOs providing legal support against the government —domestically and internationally.
Surveillance of Romanian Journalists Threatens Democracy: On March 17, 2025, journalists Victor Ilie and Luiza Vasiliu reported they were spied on by an operative team for several days in 2023. Romanian authorities have issued surveillance warrants targeting at least three journalists, including Rareș Neamțu and Tudor Leahu from 7 Iași, who have been subjected to repeated surveillance measures.
Ukraine Ministry of Culture Suspends Exit Permit System for Male Journalists: On March 4, 2025, Ukraine’s Ministry of Culture announced the suspension of the exit permit system that required male journalists to obtain special authorization to leave the country that restricted journalists’ movement and international reporting.
EU Explores Funding Options for Radio Free Europe After U.S. Cuts: Following the U.S. administration’s cessation of funding to Radio Free Europe, the Czech Republic, who hosts RFE’s headquarters in Prague, is spearheading efforts to secure EU support for the broadcaster. EU foreign policy chief Kaja Kallas expressed regret over the U.S. decision.
Protest Watch:
Columbia University Considers Compliance Amid Federal Funding Threats: The Trump administration has revoked $400 million in federal funding for Columbia University due to alleged antisemitism on campus.
The administration’s demands include placing the Middle Eastern, South Asian, and African Studies Department under “academic receivership,” banning masks on campus and granting campus security arrest powers.
On Wednesday, Columbia University signaled it would comply with the administration's demands in return for restoring $400 million in federal funding, sparking debates over institutional autonomy and academic freedom.
DHS Official Suggests Protest Could Justify Deportation:
In an interview, Department of Homeland Security Deputy Secretary Troy Edgar repeatedly avoided answering whether protesting is a deportable offense, instead implying that foreign nationals who engage in demonstrations might not have been allowed into the U.S. in the first place.
The discussion centered around the case of Mahmoud Khalil, a lawful permanent resident married to a U.S. citizen who was arrested and faces deportation.
Edgar suggested that if Khalil had stated his intent to protest on his visa application, he may not have been admitted.
When pressed on whether protesting is considered terrorism, Edgar responded: “Had he watched it on TV? It’s pretty clear.”
Why it Matters: The conflation of protests with criminal acts of terrorism could have severe chilling effects on speech, especially for non-citizens.
State Department Revoked Columbia University Student’s Visa: The U.S. State Department has revoked the student visa of Ranjani Srinivasan, a doctoral student at Columbia University, citing allegations of advocating for violence and terrorism. Srinivasan's attorney disputes these claims, asserting that her political speech is protected under the First Amendment.
Greenpeace Hit with $660M Verdict: A North Dakota jury ordered Greenpeace to pay over $660 million in damages to Energy Transfer for its role in protests against the Dakota Access Pipeline. Greenpeace condemned the ruling as a corporate attempt to silence dissent, labeling it a SLAPP (Strategic Lawsuit Against Public Participation). The group plans to appeal.
Istanbul Protests Banned, Social Media Restricted in Response to Mayor’s Arrest: Ekrem İmamoğlu, the mayor of Istanbul and a prominent challenger to President Recep Tayyip Erdoğan, was detained by Turkish authorities on charges of corruption and alleged ties to a terrorist organization. Istanbul University annulled İmamoğlu's degree, which could bar him from contesting in the upcoming presidential elections. Authorities imposed a four-day ban on demonstrations in Istanbul and restricted access to major social media platforms, including X (formerly Twitter), YouTube, Instagram, and TikTok. So far, they have detained 37 individuals for “provocative” social media posts following the arrest.
Outrage After Mozambique Security Forces Shoot Protester: On March 5, 2025, 31-year-old designer Alcina Nhaume was shot in the face inside her atelier in Mozambique by security forces. The incident occurred during a security operation in the capital when security forces deployed to disperse protests opened fire.
Georgia Authorities Freeze Accounts of Organizations Supporting Protestors: On March 17, 2025, Georgian authorities froze the bank accounts of the 5 civil society organizations, alleging their involvement in “sabotage” related to ongoing anti-government protests. The Prosecutor’s Office claims these organizations provided financial assistance to individuals fined or detained during protests, which organization’s defend aligns with their human rights mission.
Uganda Arrests During Patriotism Program Raise Free Speech Concerns: On March 18, 2025, three men were arrested in Bugiri District for allegedly disrupting a government-led patriotism training at Bukooli College. The individuals reportedly shouted accusations of the event serving as a front for the National Resistance Movement to promote President Museveni’s extended rule.
Quick Hits:
Global Survey Reveals Declining Support for Free Speech: A new report from The Future of Free Speech analyzes global attitudes toward free expression, revealing declining support for free speech protections in multiple regions.
Younger generations express greater support for restricting offensive speech, particularly regarding hate speech and misinformation.
Support for press freedom varies widely, with strong backing in Europe but waning support in parts of Asia and the U.S.
While many respondents endorse academic freedom, opinions shift when controversial topics arise, highlighting increased polarization on university campuses.
The report also finds that people are less likely to allow AI-generated content on controversial topics.
Spanish Supreme Court Upholds Fine for Offensive Tweet Against King Felipe VI: The Spanish Supreme Court has confirmed a €720 fine imposed on Albert B.T. for a tweet posted on March 18, 2020, during King Felipe VI's televised address on the COVID-19 pandemic. The tweet stated: "Let's cut the neck of this son of a bitch, we're taking too long."
Hungary Bans Pride Events: On March 18, 2024, Hungary’s parliament passed a law banning pride events nationwide and permitting authorities to use facial recognition technology to identify and fine attendees up to 200,000 Hungarian forints.
EU Bans Huawei Lobbyists After Bribery Scandal: Belgian federal prosecutors have arrested several individuals in a corruption probe involving European Parliament members and Chinese tech giant Huawei. In response, the European Parliament has banned Huawei lobbyists from accessing its premises.
NetChoice Sues Louisiana Over Social Media Age Verification Law: On March 19, 2025, NetChoice filed a federal lawsuit challenging Louisiana's social media age verification law, Senate Bill 162, which has been in effect since July of last year. The law mandates parental permission for users under 16 and requires platforms to verify user ages.
How much do people support free speech around the world? See where your country ranks: