The rapid evolution of artificial intelligence is fundamentally reshaping our world, bringing both remarkable promise and harrowing peril. While AI offers innovative avenues to streamline humanitarian aid and solve complex logistical challenges, the UN Refugee Agency (UNHCR) has issued a sobering warning: this same technology is being weaponized to accelerate the spread of misinformation, hate speech, and hyper-realistic deepfakes. For the millions of people who have been forced to flee their homes, the digital landscape is becoming increasingly hostile. As these harmful narratives proliferate, they don’t just exist as abstract concepts on a screen; they translate into real-world dangers that jeopardize the safety of refugees and the dedicated humanitarian workers striving to protect them.
At the recent AI for Good Summit in Geneva, the UNHCR made a compelling argument that vulnerable, displaced populations must be at the forefront of international discussions regarding AI governance. Too often, policy frameworks are designed with high-resource nations in mind, leaving the most marginalized communities to navigate the digital storm without adequate protection. The agency is calling for robust, global cooperation that includes technology companies, governments, and researchers to create safeguards specifically tailored to humanitarian contexts. This isn’t just about tweaking algorithms; it is about ensuring that content moderation is effective in less widely spoken languages and that digital safety measures are robust enough to defend those whose voices are most easily silenced or misrepresented.
The human cost of this digital disinformation is staggering, as online toxicity often bleeds into physical violence and civil unrest. The UNHCR reports that dehumanizing narratives and false claims have moved far beyond the digital sphere, fueling attacks on refugees, inciting protests, and systematically blocking access to essential services like housing, education, and employment. In some of the most extreme cases, this artificial friction has led to secondary displacement and the tragic loss of life. By fostering an environment of suspicion and resentment, bad actors are effectively turning host communities against those who are already at their most vulnerable, significantly hindering the prospects of successful integration and peaceful coexistence.
Particular alarm has been raised over the increasing use of generative AI to create deceptive and abusive content. Recent incidents in Libya serve as a harrowing case study, where deepfake videos—some pretending to be UNHCR representatives—have been used to sow chaos and distrust. Furthermore, digital platforms have been exploited to reveal the precise locations of humanitarian staff, putting lives at risk, while messages filled with inflammatory rhetoric have been mass-produced to incite physical violence. For groups like the Rohingya, who have already endured horrific persecution, these persistent, automated hate campaigns follow them across borders, creating a cycle of trauma that technology is currently exacerbating rather than alleviating.
The internal data from the UNHCR paints a grim picture, with 93 percent of humanitarian staff reporting that they have witnessed how digital misinformation directly undermines their critical missions. This impact is disproportionately felt by women, who face a rising tide of gendered harassment and abuse, made easier and more scalable by generative AI tools. Beyond the immediate threat of violence, this digital chaos also serves as a potent weapon for smugglers and human traffickers. By using AI to create elaborate, fake promises of safe passage, jobs, or legal pathways, these criminal networks lure desperate individuals into life-threatening situations. Preventing this exploitation is not about stifling free expression; it is about defending the basic human right to safety and integrity of information.
As we look toward the future, the UNHCR is taking proactive steps to combat these threats, most notably through the establishment of a “Community of Practice on Information Integrity in Humanitarian Contexts.” This initiative, alongside a practical toolkit shared with thousands of humanitarian practitioners, helps local teams identify risks and respond to misinformation before it spirals out of control. However, the agency acknowledges that this is not a problem they can solve alone. The protection of refugees in the digital age now depends on a new, urgent social contract where technology companies and governments treat humanitarian safety as a core design principle rather than an afterthought. If we are to harness AI for the “good” of humanity, we must first ensure it does not become a tool for shattering the lives of those who have already lost everything.