In our rapidly digitizing world, the line between online discourse and physical safety has become alarmingly thin. UNHCR, the UN Refugee Agency, recently highlighted a troubling trend: the dangerous surge of misinformation, hate speech, and deepfakes is no longer just a digital nuisance; it is causing real-world, life-threatening harm to refugees and those who strive to help them. While Artificial Intelligence (AI) is often praised for its potential to revolutionize industries, it is simultaneously destabilizing our shared information ecosystem. By accelerating the spread of deceptive content, AI is eroding the very foundation of trust required to protect the world’s most vulnerable people. However, this technology is not inherently malicious; it represents a double-edged sword that, if governed with empathy and inclusivity, could also serve as a powerful tool to provide stability and support in humanitarian crises.
At the recent AI for Good Summit in Geneva, leaders from government, technology, and academia gathered to confront this precarious reality. The core message was clear: humanitarian perspectives must be woven into the fabric of global AI governance. It is not enough to design global tech policies in a vacuum. We need content moderation tools that are robust enough to function effectively in humanitarian contexts and accessible in the less-common languages spoken by displaced populations. This requires putting human beings—specifically, people who understand the nuanced realities of displacement—at the heart of safety teams. Without clear guardrails to prevent the weaponization of AI, we risk allowing the technology to exacerbate the already dire conditions faced by refugees across the globe.
The scale of this issue is immense, as the world’s largest displacement crises are frequently accompanied by “information crises.” From Africa to the Asia-Pacific region, we are witnessing how distorted information acts as a barrier to basic human rights. When rumors, scapegoating, and dehumanizing rhetoric spread online, the consequences are felt in the streets, the job market, and the classroom. These digital attacks degrade social cohesion, making it harder for refugees to integrate and find safety. In the most tragic instances, online vitriol has been the direct precursor to physical violence, protests, and even killings. For the displaced, the internet—a place meant to offer connection—has become a constant source of existential threat.
The impact is felt perhaps most acutely by UNHCR staff, who are seeing their ability to deliver life-saving services compromised by hostile narratives. Surveys indicate that a staggering 93% of UNHCR employees have witnessed the toxic effects of disinformation, a problem that disproportionately targets women. The advent of generative AI has escalated this significantly, with deepfake videos impersonating officials and staff designed to sow confusion and panic. When refugees, already traumatized and exhausted from their journeys, are targeted by these sophisticated deceptions, they lose access to legal pathways and life-saving assistance. We have already seen this manifest in places like Libya, where dangerous misinformation turned the tide of public sentiment, putting both refugees and the aid workers trying to protect them in mortal danger.
This crisis of integrity knows no borders. Dehumanizing narratives often follow refugees into exile, preventing them from feeling secure even long after they have fled their homes. While UNHCR staunchly defends the fundamental right to freedom of expression, there is a vital distinction to be made between respectful public debate and the organized malice of bad actors. Smugglers and traffickers, for instance, are increasingly using digital platforms to feed false promises to desperate families, leading them into exploitative and dangerous situations. Protecting refugees means ensuring they have access to truthful, reliable, and life-saving information, which is a necessary component of their protection and safety.
The road ahead requires a collaborative, hands-on approach. No single agency or nation can defeat the tide of misinformation alone. UNHCR is now spearheading efforts like the “Community of Practice on Information Integrity,” a direct partnership between governments, tech giants like Google, and local practitioners, to systematize our response. From the “Information Integrity Response Toolkit” to better localized risk assessments, the goal is to shift from reactive damage control to proactive community support. Ultimately, as we shape the future of AI governance, we must ensure the voices of those who have been marginalized are not left behind. By prioritizing trust and human dignity, we can turn the tide and ensure that the digital age becomes a beacon of safety for those who need it most.