When the Qilin group claimed an attack on subsidiaries of Japan’s Keio Railway in late September, it published 27 images it said showed internal documents and set a countdown for releasing more. Keio said on October 16 that it was still checking whether the material was genuine. The tactic, theft plus a public clock, is now routine, and a new global survey suggests the pressure on victims is growing in two ways: more AI behind the attacks, and more threats against the people who work at targeted firms.
What the survey found
The 2026 Ransomware Risk Report from Semperis polled 1,200 IT and security professionals across nine countries and eight industry sectors. Among victims, 84% said attackers had exploited AI in their incidents, and 44% said the attackers had turned the victims’ own AI systems against them. Sixty-nine percent had suffered multiple successful attacks within the previous year, and 59% of respondents had paid a ransom at least once. Attacks on identity systems, the infrastructure that controls who can enter a corporate network, hit 87%.
The human side is just as stark. More than a third of victims faced psychological or physical threats against employees. In 14% of organizations, staff or family members were forced to change jobs, office locations or homes because of sustained threats.
Payments are down, but not the pressure
Ransomware payments reached a record $1.25 billion in 2023 and fell to $813 million in 2024, a drop attributed to victims refusing to pay and to law enforcement action. The survey shows the attacks themselves have not eased.
Why AI changes the equation
Chris Inglis, a former White House National Cyber Director who sits on the Semperis board, called AI an inflection point. In his view it lets attackers spot weaknesses faster, tailor social engineering to specific employees, automate an intrusion and adjust as defenders react. Someone without technical skills could ask an AI system to write the code and target a company. “I think we’re already seeing autonomous attacks,” he said. “Autonomous ransomware attacks are included in the mix.”
Tomer Bar, Semperis’s associate vice president of security research, warned that attribution will get harder for law enforcement as development speeds up. Attackers can also test their code with large language models, so flaws in their encryption are less likely.
When a firm’s own AI turns on it
Courtney Guss, the firm’s associate vice president of crisis management, cautioned that AI agents sometimes seek extra permissions to finish a task, and the quickest route is to compromise the identity system. An attacker who manipulates one agent, or the credentials it uses, gains another way deeper into a network.
Inglis offered a counterweight. Many attackers already live off the land, using whatever tools a victim has, and if AI is one of them they will use it. But the weakness that let them in is probably the same one as before, a failure to control access to the network. His advice: “Flip the script. You must use AI in order to counter that threat.”
Sources
- Qilin Threatens to Release Data Stolen in Keio Railway Cyberattack
- Ransomware
- AI linked to 84% of ransomware attacks, new report finds
Photo: Tamatak 'v chamatak ki (Public domain)
