Misinformation and hate speech are inciting harm to refugees, with artificial intelligence exacerbating the spread, the United Nations has warned.

The UN’s refugee agency UNHCR has urged tech giants to help turn the tide and help put AI to good use in managing humanitarian crises.

Speaking at the AI for Good Global Summit in Geneva, UNHCR information integrity chief Gisella Lomax said the world’s major displacement crises are often twinned with “information crises”.

“The spread of misinformation, hate speech and deepfakes is exacerbating and inciting real-world harm to refugees and humanitarians,” Ms Lomax said.

She this swirl of misinformation and disinformation had resulted in an increase in “false, hostile narratives targeting operations and staff.”

“In war and conflict and emergencies, these information risks are extremely sharp,” and can quickly lead to harm, Ms Lomax said.

She said trusted information was a necessity for refugee protection.

“We need to turn the tide on this, and I do see some encouraging progress,” she said, adding that it needed to be systematised and scaled up.

“Our call to action is to all tech companies, AI companies and digital platforms to partner, invest and collaborate with humanitarian organisations.”

Ms Lomax said that when information was distorted, it could reduce access to jobs and education, make integration harder and threaten social cohesion.

“Online rumours, false accusations, scapegoating and dehumanizing speech have triggered protests and attacks. In extreme cases, they have been linked to physical violence, killings, and are even a factor in forced displacement,” she said.

UNHCR has previously said generative AI was compounding the effects, at scale, saying deepfake videos of its staff and refugees were presenting a growing challenge.

“This isn’t about how legitimate societal concerns on migration and asylum are debated online – but about life-threatening information risks that can lead to serious harm in humanitarian contexts,” Ms Lomax said.

She gave examples such as smugglers and traffickers exploiting digital platforms to deceive people with false promises of safety and employment, leading them into dangerous situations.

Ms Lomax said UNHCR wanted to ensure that content moderation tools work in humanitarian contexts, and in less-common languages.

The agency said there were 117.8 million forcibly displaced people worldwide at the end of 2025, of whom 68.7 million were internally displaced and 35.6 million were refugees under its mandate.

Of those refugees, two-thirds come from five countries: Venezuela, Ukraine, Syria, Afghanistan and Sudan.

The Geneva summit is an attempt by the wider UN to focus on using AI’s potential to serve humanity by solving global challenges, and to look at the state of AI standards.