Multinational organisations are increasingly turning to artificial intelligence (AI) to verify risk information, but misinformation, deepfakes and slow decision-making continue to create significant challenges during crises.
This is according to a report from International SOS, which warned that AI-generated deepfakes were the “new frontier of biowarfare”.
Four in 10 organisations responding for the research said their organisation had either been affected by, or lacked sufficient visibility, to determine the impact of false, misleading or unverified information related to health, wellbeing or security risks.
Respondents cited fake signals creating operational uncertainty, synthetic content increasing credibility risks, and targeted deception and impersonation.
However, nearly 60% of respondents viewed the greatest potential for AI in verifying or validating information across multiple sources.
Broader structural challenges
The Risk Outlook 2026 Pulse Check report, based on responses from 257 senior risk, security, and health professionals across 53 countries, also indicated broader structural challenges in organisational risk response, with clarity of responsibility and decision-making processes emerging as significant concerns.
Only 10% of organisations said they could respond “very quickly” to new risks. While 68% believed they responded “fairly quickly”, this may not be sufficient in an increasingly volatile operating environment.
One in five organisations (20%) admitted they were “not quick to respond at all”.
But the research also exposed uncertainty within risk governance structures, with 17% unclear about what happens next when new risks emerge, 25% unclear about escalation triggers, and only around 20% being very clear on how global and local responses aligned.
‘New frontier of biowarfare’
Irene Lai, global medical director at International SOS, said: “AI-generated deepfakes of real health professionals stating false information on measures to contain a natural outbreak is the new frontier of biowarfare, weaponised from a computer screen, without the need for introduction of a biological agent.”
Cvete Koneska, global security director at International SOS, said: “The potential is real, but so too is the risk of overestimating the maturity of current tools.
“While AI can help surface, synthesise and prioritise information at speed, verification remains a high-stakes task where careful human judgement remains indispensable, particularly when decisions can impact people’s health, safety, and security.
“Misplaced reliance on AI at this stage can amplify misinformation risks and delay critical decisions rather than improve them. The most prudent approach is a human-led model of integrating AI into risk management.”
