It was a moment of quiet alarm in the middle of a busy diplomatic briefing. Speaking at the Palais des Nations in Geneva, UNHCR’s Senior Advisor on Information Integrity, Gisella Lomax, laid out a reality that is not yet fully understood by much of the world: the digital information landscape is no longer just an abstract space of opinions and debates, but a battlefield where the lives and safety of refugees and humanitarians are increasingly at stake. The UN Refugee Agency is growing deeply concerned about the spread of misinformation, hate speech and sophisticated deepfakes, and how these are not just polluting conversations online, but are actively causing real-world harm. People who have already lost their homes, their communities and their sense of safety are now being met with another wave of danger, one that travels through screens and spreads at the speed of a click. Artificial intelligence is making this worse, supercharging lies, distortions and malicious narratives with terrifying scale and speed, and eroding the trust that humanitarian work depends upon. Yet Lomax made it clear that technology is not simply an enemy. If harnessed with intention, thoughtfulness and inclusion, AI could also become part of the solution, helping to identify threats, protect the vulnerable and ensure that truth can still find a way through the noise.
At the AI for Good Summit in Geneva, UNHCR brought together governments, technology pioneers and academic experts to wrestle with these complicated issues. The goal was not to reject technology, but to shape it in ways that genuinely protect human dignity. Lomax stressed that it is absolutely critical for refugee and humanitarian perspectives to be woven into the global conversation about AI governance, because the people who are most affected by digital manipulation are rarely in the rooms where those decisions are made. For example, content moderation tools are often built around the world’s dominant languages and cultural contexts, leaving them weak or ineffective in the places and languages where humanitarian work actually unfolds. A platform designed to catch hate speech might not understand the nuance of a local dialect or a particular refugee context. That is why UNHCR is pushing for a shift: trust and safety teams need to be deeply connected to the human realities on the ground, and clear frameworks need to be established around guardrails, manipulation and abuse. This is no longer a technical concern; it is a life-and-death priority. From Africa to the Asia-Pacific, the world’s largest displacement crises are almost always information crises as well. When lies take root, they change behaviour, they shape policies, they ignite fear, and they place already traumatized people at even greater risk.
The human consequences are impossible to overstate. Misinformation reduces refugees’ ability to access jobs and education, it makes integration vastly harder, and it eats away at the social cohesion between displaced people and the communities hosting them. Online rumours, false accusations, scapegoating and dehumanizing speech have led directly to protests and physical attacks. In the most extreme cases, they have been linked to killings and have even forced people to leave new places of safety, creating secondary displacement. Lomax noted that even UNHCR’s own digital communications, specifically those for the International Day to Counter Hate Speech, were flooded with hateful comments only last month. That shows how easily the toxic conversation can settle into official spaces. The spiral of mis- and disinformation is also becoming a serious threat to humanitarian operations and staff security. There has been a disturbing increase in false and hostile narratives targeting UNHCR operations and its people. These narratives can undermine fundraising at a time when budgets are shrinking and needs are growing, which means that some of the world’s most vulnerable people may end up paying the price with their access to shelter, food and legal protection. A recent UNHCR survey found that 93 per cent of staff respondents had personally witnessed misinformation, disinformation or hate speech affecting the delivery of the agency’s mandate. Women, both among refugees and among staff, are disproportionately targeted, and generative AI has magnified this abuse, allowing attackers to create damaging deepfake videos of UNHCR staff and refugees with astonishing ease.
Nowhere is this crisis more vivid than in Libya, where a recent wave of violent hate speech and dangerous misinformation actively incited hostility and abuse against refugees. The safety of UNHCR staff and other humanitarian workers was affected too. The tactics are ugly and ruthless: deepfakes of the UNHCR Representative have circulated; people are being encouraged online to share the GPS coordinates of staff addresses; and videos have emerged describing national staff as traitors to their own country. These are not just unkind comments. They are weapons aimed at the most basic structures of human dignity and security. And hate speech does not respect borders. The Rohingya, for example, were forced to flee violence and persecution, yet they continue to face dehumanizing narratives online, even in exile. This online hate translates into harassment and intimidation in the real world, making it harder for refugees to rebuild their lives and deepening divisions between displaced families and host communities. Lomax was careful to underline that UNHCR fully upholds freedom of expression as a fundamental human right. Everyone, including refugees, has the right to receive and share ideas and information. The concern is not about legitimate societal debates over migration and asylum, which should happen openly and honestly. The concern is with information risks that can literally lead to serious harm in humanitarian contexts. Smugglers and traffickers, for example, are exploiting digital platforms with clever disinformation, luring people with false promises of safety, legal pathways and employment, only to push them into dreadful, exploitative situations.
Faced with such a daunting challenge, UNHCR has chosen a practical and collaborative response. With the support of Switzerland, the agency created a Community of Practice on Information Integrity in Humanitarian Contexts, bringing together partners at both local and global levels to share lessons and strengthen responses. Another important step is the Global Refugee Forum Multistakeholder Pledge on Information Integrity, launched with Google in 2023, which aims to inspire more multilateral cooperation and encourage different actors to take responsibility for the information ecosystem. Encouraging progress is happening, but as Lomax noted, it needs to be systematized rather than remaining isolated initiatives. At the heart of all this work is the recognition that refugees themselves must be key participants, not just subjects. Local partnerships shaped UNHCR’s Information Integrity Response Toolkit, which is already being used by around 2,000 practitioners to guide practical action, from risk assessments to community-based responses. This toolkit is a visible example of what can happen when the people closest to the problem are included in building the answer. It is not enough for external experts and tech companies to swoop in with generic solutions; the work must be grounded in the lived experiences and wisdom of communities on the ground.
The challenge is fast-moving, deeply complex, and no single actor can solve it on their own. But the message from UNHCR is both urgent and hopeful. Refugee and humanitarian contexts must not be left behind in the conversations about AI governance, because the future of that governance will shape the safety of millions of people. Trusted information is not a luxury or an afterthought; it is a necessity for refugee protection. Information integrity helps save lives, preserve dignity, and maintain the fragile trust that allows humanitarian work to happen at all. Lomax’s briefing was a reminder that in a world where reality can be manipulated, the simplest human connection becomes heroic. A refusal to lie, an insistence on protecting the vulnerable from hatred, a commitment to ensuring that those who have lost everything are not also stripped of the truth, all of this is humanitarian work in the deepest sense. The digital world is not separate from the physical world. It is the place where stories are told, where fears are shaped, and where communities either find each other or turn against each other. If we can build a better information ecosystem, one rooted in inclusion and integrity, then we can give refugees a fairer chance at safety, and give the world a stronger foundation for peace. But to do that, everyone must step up, not just governments and aid agencies, but every person who clicks, shares, and speaks.