Semperis finds 44% of surveyed ransomware victims said their own AI was exploited

Companies’ own AI systems are becoming tools for ransomware attackers, according to Semperis’s newly released 2026 Ransomware Risk Report. CBS News reported that 44% of surveyed victims said attackers exploited their AI agents or systems. Yet Chris Inglis, the former U.S. National Cyber Director affiliated with Semperis, argues that companies should use AI to defend themselves—not abandon it.

A victim survey, with conflicting AI-use figures

Semperis, a cybersecurity company focused on identity protection and recovery, published the findings on October 14, 2026. Its 2026 Ransomware Risk Report surveyed 1,200 IT and security professionals across nine countries and eight industry sectors. The company said only organizations victimized by ransomware gangs in the preceding 12 months were included.

That scope is important: the results describe experiences reported by ransomware victims, not the likelihood that any company will suffer an attack. They also represent respondents’ accounts of AI involvement, rather than independently verified measurements of attackers’ tools. The survey’s reach should not be confused with a census of ransomware activity.

There is also an unresolved numerical discrepancy. CBS News’s account puts attackers’ use of AI at 84% of surveyed victims. Semperis’s own release says threat actors weaponized AI in 87% of attacks. Those figures should not be treated as interchangeable or as independent corroboration; both describe findings from the same named report.

An AI tool can become an attacker’s tool

The finding about companies’ AI systems is different from criminals using AI to write malicious code. An AI agent is software that acts across applications for a user. It may have access to sensitive corporate systems. Manipulating the agent or stealing its credentials could give attackers another way into a network.

Courtney Guss, Semperis’s associate vice president of crisis management, warns that agents seeking more permissions could compromise identity systems: the infrastructure that determines who can access a company’s network. Her concern is that an agent pursuing its assigned task could obtain the power to grant itself further access, even without malicious intent.

Inglis offers a caution against treating AI adoption itself as the cause of a breach. Attackers already use tools and data they find inside a victim’s infrastructure to blend in and move around, he told CBS News. If AI is among those tools, they may use it, while the weakness that enabled entry may predate the AI deployment.

He nevertheless sees AI increasing attackers’ capacity. Inglis said it can help identify weaknesses, tailor deceptive approaches to employees, automate intrusions and adjust tactics as defenders respond. He also described how someone without sophisticated technical knowledge could ask AI to write attack code. These are his assessments of AI’s capabilities, not separate measurements from the survey.

Inglis, whom CBS identifies as a Semperis board member, said he believes autonomous attacks are already occurring. His view goes beyond the survey’s broad finding of AI involvement: using AI during an attack is not, by itself, evidence that the attack ran autonomously.

I think we're already seeing autonomous attacks. Autonomous ransomware attacks are included in the mix.

Chris Inglis, former U.S. National Cyber Director and Semperis board member, speaking to CBS News

Repeat attacks, personal threats and recovery gaps

Ransomware’s effects extend beyond how attackers break in. CBS reported that 69% of surveyed victims faced multiple successful attacks in the previous year, while 59% of respondents paid a ransom at least once. Semperis separately reported that 31% made multiple payments to attackers.

Semperis says 40% of attacks involved psychological or physical threats against staff. CBS reported that 14% of surveyed organizations said employees or family members had to change jobs, office locations or residences because of sustained ransomware threats. The pressure described here reaches beyond encrypted data and disrupted business operations into employees’ personal lives.

Meanwhile, Semperis says 87% of ransomware attacks targeted identity systems, including Active Directory, Entra ID and Okta. Its findings describe gaps in preparation for restoring those systems after an attack:

• Half of surveyed companies maintained dedicated identity backup systems.
• Fewer than half automated identity-system recovery.
• Forty percent had disaster recovery or crisis response plans that omitted Active Directory recovery procedures.