AI Fuels More Ransomware Attacks as Payments Decline

Introduction
Ransomware activity is climbing again, and artificial intelligence is doing a meaningful share of the groundwork. The source material behind this report states that large language models and other AI systems are enabling a growing number of ransomware groups to craft convincing phishing emails and to locate sensitive data far more quickly than manual effort allows. The same material notes a counterintuitive twist: even as attack volume rises, ransom payments are declining. That combination matters because it breaks the long-held assumption that more intrusions automatically translate into bigger payouts, and it forces defenders to reconsider where the real risk now sits.
For IT and security teams, the practical takeaway is not that ransomware has become less dangerous. It is that the economics of extortion are shifting underneath a rising tide of attacks, and the defensive controls that once worked against sloppy, template-driven campaigns are no longer sufficient.
How AI lowers the bar for ransomware operators
Two capabilities stand out in the source material, and both compress work that used to require skilled humans.
The first is phishing quality. Historically, defenders leaned on the tell-tale signs of a rushed, poorly translated lure: awkward grammar, generic greetings, implausible context. Language models remove those signals. Operators can generate fluent, contextually appropriate messages at volume, adapt tone to a specific industry or even a specific recipient, and iterate on wording without hiring a single native speaker. The result is social engineering that reads like ordinary business correspondence, which raises the cost of relying on employee vigilance as a primary control.
The second is data discovery. Ransomware extortion depends on finding the files that hurt: customer records, financial data, intellectual property, legal correspondence. AI systems can help intruders triage large volumes of unstructured data quickly, identifying what is sensitive and therefore monetisable. That shortens the gap between initial access and a credible threat to leak, which is precisely the window defenders try to exploit.
The source material does not name specific ransomware operations, quantify the increase, or attribute campaigns to particular tooling, and those details are not yet public. What it does establish is direction: more groups are adopting these techniques, not fewer.
Why payments are falling while attacks rise
The decline in payments alongside a rise in attacks is the more interesting half of the story, because it suggests victims are getting harder to monetise even as they become easier to reach. Several dynamics plausibly explain the divergence, though the source material does not quantify them:
- Better recovery posture. Organisations that can restore from tested, isolated backups have far less incentive to pay, and that message has been repeated by incident responders for years.
- Legal and sanctions exposure. Paying a sanctioned entity carries real regulatory risk, and legal counsel increasingly sits in the decision path before any transfer is approved.
- Law enforcement pressure. Disruption operations, takedowns and public advisories raise the perceived risk on the attacker side and reduce victim confidence that payment solves anything.
- Reputational and ethical calculus. Boards are more willing to accept operational pain than to fund criminal infrastructure, particularly where payment offers no guarantee that stolen data is deleted.
- Insurance scrutiny. Insurers now ask harder questions about controls before and after an incident, which changes the financial logic of capitulating quickly.
None of this means extortion has stopped working. It means the return per victim is compressing, which historically pushes criminal groups toward volume, toward smaller and less prepared targets, and toward data-theft-only extortion that does not require encrypting anything at all.
Who is actually exposed
The organisations most affected by this shift are not necessarily the largest. When phishing quality and data triage become cheap, the minimum viable target shrinks. Mid-sized businesses, professional services firms, healthcare providers, manufacturers and public-sector bodies that previously fell below the threshold of interest become economically attractive because the marginal cost of attacking them has dropped. Any organisation with email-facing staff and large stores of unstructured sensitive data should assume it is now inside the addressable market.
What defenders should do now
Because the attacker's cost curve has fallen, defensive strategy has to shift from filtering obviously bad messages to assuming that some convincing ones will get through.
- Stop depending on user detection. Deploy phishing-resistant multi-factor authentication, enforce DMARC, tag external senders, and isolate links and attachments at the gateway.
- Harden identity. Restrict conditional access, block legacy authentication protocols, and monitor for session token theft, which bypasses password and MFA controls entirely.
- Know where your sensitive data lives. Classify it, reduce duplication, apply least privilege, and alert on bulk access or unusual collection activity that precedes exfiltration.
- Shorten detection time. AI-assisted triage cuts attacker dwell time, so defenders need equivalent speed: centralised logging, behavioural analytics on file access, and alerting on large outbound transfers.
- Test recovery, not just backups. Immutable, offline copies plus regular restore drills are what actually remove the incentive to pay.
- Pre-agree the extortion decision. Legal, insurance, communications and executive leadership should have a framework in place before an incident, not during one.
- Run AI-realistic tabletops. Exercise against high-quality, well-written lures rather than obviously fake ones, and test how quickly the organisation detects a fast data-grab scenario.
Grivyonx Expert Analysis
The headline reads like good news, and it is not. Falling payments do not signal a weakening threat; they signal a threat that is being forced to change shape. When AI drives the cost of a credible phishing campaign toward zero, attackers respond to lower yields per victim by attacking more victims, and by shifting from encryption-based extortion to pure data theft where the overhead is lower. Defenders who read declining payments as a reason to relax will be the ones who fund the next quarter's ransomware revenue. The correct interpretation is that the volume of attempts is rising, the quality of each attempt is rising, and the only durable counterweight is reducing the value of what an intruder can reach and the speed at which they can reach it.
Conclusion
AI has not invented a new kind of ransomware, but it has industrialised the parts that used to be bottlenecks: writing persuasive lures and finding the data worth stealing. The simultaneous decline in payments suggests victims and their advisers are pushing back more effectively, which is genuinely encouraging, but it also predicts more attempts against more organisations rather than fewer. The organisations that come through this period intact will be those that treat phishing-resistant identity, data minimisation and rehearsed recovery as baseline hygiene rather than projects. Teams looking to operationalise that work, from AI-driven detection and response automation to secure pipelines and cloud hardening, can find support through Grivyonx's AI automation, DevSecOps and cloud security services.
- #ransomware
- #AI security
- #phishing
- #cybercrime
- #incident response
- #threat intelligence
- #DevSecOps



