Welcome to the tenth edition of HRF’s AI for Individual Rights newsletter.
We start in China, where AI-powered tools are censoring teachers in universities. From there, we travel to Vietnam, where the ruling Communist Party is preparing to use AI to flood the internet with pro-regime propaganda and remove content that violates Party guidelines.
We then revisit the 2026 Oslo Freedom Forum Freedom Tech Track, where more than 600 attendees examined AI’s growing role in authoritarian repression while exploring how these same technologies can be used to advance freedom. You can watch the full livestream recording here. Alongside the track, HRF launched the AI Lounge, an interactive space where attendees could experiment with cutting-edge AI agents and open-source tools. For many, it was a glimpse into a future where activists wield capabilities once available only to governments, large corporations, and well-funded institutions.
We close with an interview from Andrej Karpathy, one of the world’s leading AI researchers (and who recently joined Anthropic). He explains how AI agents are opening new possibilities for software development and how they dramatically expand what a single person can create. HRF is excited about what this could mean for nondevelopers seeking to build tools and expand their work despite limited resources. If you have some time today, this is worth your while. Watch it here.
HRF’s AI Agent Camps: Supercharging Activism with AI
Over the past 18 months, HRF has expanded the AI for Individual Rights Program through training, education, research, events, and research. One of our proudest achievements is launching “Agent Camp,” a hands-on training program designed to help human rights defenders deploy open-source AI agents. The video below captures the energy, collaboration, and real‑world impact that AI is having on activism. Watch it now to see how AI is democratizing access to innovation and powering global movements.
The Latest in AI for Repression
AI-Powered Surveillance Flags Sensitive Topics in Chinese Universities
An article published by Perpetual Light Studio, a Chinese journalism outlet, and translated in the ChinAI newsletter revealed that Chinese universities are turning classrooms into AI-powered surveillance spaces. Cameras now track students’ attention and behaviors, while also examining professors’ speech, gestures, and whether lectures trigger “sensitive keywords.” A lecturer of ideological and political education classes, under the pseudonym Xiaoxi, argues the system has made it harder to speak freely. “I originally aspired for a truly brilliant classroom environment; yet, suddenly finding myself under constant surveillance, I seem to have shifted toward pursuing the avoidance of errors and compliance,” said Xiaoxi.
Why this matters: Open criticism of the Chinese Communist Party (CCP) is already dangerous and a punishable crime in China. AI surveillance in universities adds another layer of surveillance and punishment. Professors are cornered into self-censorship. Students are losing access to honest discussion. And classrooms are being redesigned to serve authoritarian propaganda rather than challenge it.
Vietnam to Implement AI for Propaganda and Censorship
Vietnam’s ruling Communist Party plans to expand their propaganda and censorship campaigns with AI. A strategy under consideration would recruit at least 5,000 AI specialists and 1,000 influencers by 2030 to saturate the Vietnamese-language internet with pro-regime content. Podcasts, short videos, and targeted social media content would explain party policies. Vietnamese tech companies, already subject to strict controls over online speech, would also develop AI tools to help “lead social discussion,” likely in the regime’s favor. The strategy would also extend to censorship, calling for AI tools to be used to remove at least 90% of content that infringes party guidelines.
Why this matters: Such a plan could drown out independent journalism and dissenting voices, making free expression increasingly difficult.
China Developing AI-Powered Tools to Predict Dissent
Leaked documents revealed that Chinese technology company, Geedge Networks, is developing AI-powered tools aimed at predicting who might one day criticize the CCP. Working with MESA Lab, a regime-backed research lab, Geedge is reportedly building behavioral profiles of individuals based on their social media, location data, telecommunications, and online activity. AI models could then sift through profiles and spotlight those who might be a political risk. There is no evidence the technology is finalized or deployed, but if deployed, China’s surveillance system could become a tool to chill dissent before it ever begins.
United Arab Emirates Plans AI-Run Government
The United Arab Emirates (UAE) announced a plan to integrate agentic AI across half of its government operations within the next two years. Every federal employee will receive AI training, and every ministry and government entity will be evaluated on how quickly it adopts the technology. Officials say the plan will increase efficiency, but critics warn that expanding AI across those systems could increase how much data is collected, stored, and analyzed. In the context of an authoritarian regime that restricts free speech and punishes dissent, this data collection could accelerate surveillance and identification of critics and could expand the regime’s control.
NVIDIA CEO Meets with Xi Jinping
Jensen Huang, the CEO of AI chipmaker NVIDIA, joined US President Donald Trump in his two-day Beijing summit with the head of the CCP, Xi Jinping. Huang’s presence signals a warning sign that China may gain access to NVIDIA’s powerful AI chips. No major deal emerged from the summit, but these friendlier relations could give Beijing a clearer path to hardware or corporate collaboration it needs to supercharge its AI ambitions. Most concerning, those advances could help the authoritarian regime further develop its already sophisticated AI-powered surveillance apparatus used to monitor minorities, track citizens, and crush dissent.
Why this matters: This move could also strengthen the CCP’s global AI influence. China currently leads the world in open-weight AI models because their models are cheap and powerful despite its lack of access to advanced AI chips. With more powerful AI chips, these systems could become more capable and deepen global reliance on AI models that are developed under CCP control and shaped by its propaganda.
HRF-Sponsored Report Cited in Anthropic Research Paper
An HRF-sponsored research report from the Australian Strategic Policy Institute (ASPI), “The party’s AI: How China’s new AI systems are reshaping human rights,” was cited in a recent Anthropic paper on global AI leadership. Anthropic used ASPI’s report to explain how the CCP is already automating human rights abuses with AI. The regime uses AI systems to censor speech, surveil ethnic minorities, collect biometric data, and monitor communications. The paper warns that advanced AI in the hands of authoritarian regimes could dramatically strengthen repression around the world. HRF is proud to support research that exposes these dangers and shapes the global conversation on AI and human rights.
Recommended Content
Will AI Create Permanent Dictatorships?
In this episode of the democracy podcast Civic Forum, host Rory Truex interviews two experts on China’s repressive digital practices, Margaret Roberts and Jennifer Pan. They discuss how authoritarian regimes can use AI to expand censorship, propaganda, and surveillance. But they also explain that the same technology can be used to empower dissidents and ordinary citizens. AI can help small teams identify corruption, build legal arguments, expose abuses, and ultimately hold regimes accountable. It’s worth listening to understand how AI can serve as a force of both control and liberation.
The Latest in AI for Freedom
Oslo Freedom Forum Hosted the Freedom Tech Track
On June 2, HRF hosted the Freedom Tech track at the 2026 Oslo Freedom Forum in Oslo, Norway. While continuing OFF’s long-standing focus on financial and digital repression, this year’s program placed a special emphasis on AI and its growing impact on human rights. Speakers explored how authoritarian regimes are using AI for surveillance, censorship, and propaganda, while showcasing open-source and privacy-preserving AI that can help activists and civil society without sacrificing privacy, security, or independence. We highly recommend you watch the livestream recording or the individual talks.
AI Lounge Supercharges Activists at the Oslo Freedom Forum
At this year’s Oslo Freedom Forum, HRF hosted an AI Lounge, a space where human rights defenders experimented with powerful open-source AI agents and local AI. In just two days, participants used AI agents to build more than 100 websites, including projects documenting protests in Cuba and others that expose human rights abuses in Belarus. The Lounge demonstrated that AI can help freedom fighters accomplish work that would have required far greater time, resources, and technical expertise. HRF is proud to help put these capabilities into the hands of those working to advance freedom and human rights around the world.
Apple Expanding Its Privacy-Protecting AI Capacity
Apple announced that it is expanding its Private Cloud Compute (PCC) beyond its own data centers by collaborating with Google and NVIDIA to scale capacity for this private AI inference. PCC allows Apple to handle complex AI tasks in the cloud while maintaining strong privacy protections. Under the new arrangement, Apple can leverage Google and NVIDIA’s infrastructure (which include data centers, servers, GPUs, and security chips) while retaining privacy safeguards that protect user data. The goal is to deliver more powerful AI capabilities, regardless of the underlying infrastructure, without compromising privacy.
Why this matters: This could dramatically increase the capacity of private AI and help push the AI industry toward adoption of privacy-protecting AI infrastructure and standards.
Agora Brings Censorship-Resistant Donations to Pro-Democracy Movements
At the 2026 Oslo Freedom Forum, Venezuelan opposition leader and World Liberty Congress co-founder Leopoldo López unveiled Agora: a censorship-resistant crowdfunding platform designed for pro-democracy movements. Vibecoded during AI Hack for Freedom and developed alongside Soapbox, Agora allows activists to raise funds and coordinate support without relying on platforms that governments can easily block. This is made possible by its use of Bitcoin and Nostr, two open networks resistant to centralized control and capture. The project offers a glimpse of what is possible when activists, developers, and AI come together to solve real-world problems under repression.
Maple AI Announces Personal Intelligence Platform
Maple AI, a company that provides end-to-end encrypted interactions with AI, announced the launch of a Personal Intelligence Platform — a suite of three privacy-focused AI applications. The first is Maple Research. This is a secure workspace where users can privately chat or speak to open models, upload documents and images, and even search the web. The second is Maple Proxy, an encrypted API that lets people connect other tools to Maple’s private models. Finally, the company announced it will soon launch Maple Agent, a private AI assistant with long-term memory. It will remember your preferences, track important information, and learn from the user without creating a visible record of behaviors that third parties can access. Interested users can join the waitlist here.
Why this matters: For dissidents, many AI tools could expose their sensitive data to third-party companies and leave it vulnerable to misuse by authoritarian regimes. Maple’s suite of services now gives them a way to use advanced AI to improve their workflows and receive personalized support without sacrificing privacy.
Recommended Content
From Vibe Coding to Agentic Engineering
In this interview, Andrej Karpathy, a founding member of OpenAI and now a lead researcher at Anthropic, dives into the advancements reshaping the AI ecosystem. Just a year ago, he termed the phrase “vibe coding” to describe how anyone regardless of technical experience can talk to AI in plain language and quickly build applications. Now, the capabilities have advanced further. With advancements in models and agents, developers can build software faster while still preserving security, quality, and good design. And freedom technologists can more quickly and cheaply deploy tools that counter authoritarian repression.
