Key Takeaways
- AI is helping ransomware operators work faster, but evidence does not yet point to fully autonomous attack campaigns.
- Reported ransomware victims rose 43% year over year in Q2 2026 as data theft became central to extortion.
- Businesses can reduce exposure by tightening identity controls, monitoring legitimate collaboration tools, and rehearsing incident response.
Ransomware activity has accelerated in recent months, with security researchers increasingly examining whether artificial intelligence is lowering the barrier to entry for less experienced operators. The clearest evidence so far suggests AI is acting as a productivity multiplier rather than creating an entirely new, autonomous form of cyberattack.
That distinction matters. AI can help criminals summarize stolen documents, research prospective victims, translate messages, produce convincing phishing content, and tailor extortion demands. It may also let smaller groups complete work that once required a broader team. But ransomware operations still depend on familiar ingredients: stolen credentials, vulnerable systems, weak access controls, human manipulation, and a path for extracting data.
The scale of the underlying problem is substantial. ENISA reported that ransomware accounted for 40% of financially motivated cyber events observed in 2025. The European cybersecurity agency has also identified growing malicious use of AI to enhance cybercriminal activity.
More recent victim figures point in the same direction, although they warrant careful interpretation. According to GuidePoint Security, organizations reported 2,279 ransomware victims in Q2 2026, a 43% increase from the same quarter a year earlier. GuidePoint Security also counted 91 active groups operating across 108 countries.
Those totals typically draw on public disclosures, leak sites, incident reports, and other observable signals. They do not represent a complete census. A criminal group may exaggerate its success, list the same victim more than once, or publish data from an older intrusion. Meanwhile, organizations that quietly contain an attack may not appear in public tracking at all. The direction of travel looks troubling, but precision remains elusive.
Encryption is no longer the only measure of ransomware damage. In Q2 2026, 97% of publicly disclosed attacks tracked by BlackFog involved data exfiltration. That reflects a broader shift toward double extortion, where criminals steal information before encrypting systems and then threaten to publish it. Highlighting the financial impact of this trend, Zscaler ThreatLabz recorded a 275% year-over-year increase in ransomware-related data theft, with an average ransom payment of $431,995. Even organizations with usable backups can therefore face regulatory, legal, operational, and reputational pressure.
AI fits neatly into that model. Once attackers obtain hundreds of gigabytes of corporate information, automated tools can help classify files, identify sensitive records, locate executive communications, and find material likely to increase leverage. What might previously have required days of manual review can potentially be compressed into a shorter window. The same tools can assist with drafting messages aimed at executives, employees, customers, or business partners.
Social engineering is changing too. Researchers have observed attackers abusing Microsoft Teams and Quick Assist, both legitimate enterprise products, to contact users or move further into compromised environments. AI-generated scripts and context-aware messages can make those interactions more persuasive. Would an employee question a support request that references an actual colleague, project, or internal issue? Sometimes, but the margin for error is shrinking.
The ransomware-as-a-service economy adds another layer. CISA reported that Gunra emerged in 2025 and expanded into ransomware-as-a-service operations in 2026, using double extortion. Alongside names such as LockBit and RansomHub, Gunra illustrates how developers, access brokers, affiliates, and data-leak operators can divide the work. AI may make each part more efficient without replacing the people coordinating it.
For business leaders, the practical response is less about buying a product labeled AI defense and more about strengthening controls around likely attack paths. Phishing-resistant multifactor authentication, restricted administrative privileges, segmented networks, protected backups, rapid patching, and monitoring for unusual data transfers can all reduce risk. Controls around Teams, Quick Assist, and other remote-support channels also deserve review.
Response planning matters just as much. Organizations can map responsibilities using the NIST Cybersecurity Framework 2.0 and operational practices described in CISA's #StopRansomware guidance, while testing decisions through realistic exercises. AI may be speeding up the attacker's clock. Defenders, in turn, benefit from shortening detection, containment, and executive decision cycles before an intrusion becomes a public extortion event.
⬇️