By John Dwyer, vice president; John Hendley, vice president and chief of staff; and Neil “Grifter” Wyler, vice president for defensive services, DivisionHex, Coalfire
AI is accelerating cyberattacks, expanding the range of viable targets and putting new pressure on defenders. In 2027, the organizations that stay ahead won’t be those chasing every emerging threat. They’ll be the ones strengthening identity security, accelerating response times and keeping human judgment at the center of increasingly automated defenses.
Three Coalfire cybersecurity leaders share their predictions for the year ahead and what organizations need to do now.
AI will make previously impractical targets profitable and create new opportunities for cybercrime
Cybercriminals have long favored repeatable attacks against widely used technologies. Ransomware proved the model: exploit common systems, reuse proven techniques and scale the payoff.
AI could change that equation.
Attackers may soon be able to enter unfamiliar environments, understand proprietary applications and develop just-in-time tooling to exploit them. Systems once too specialized or costly to target could suddenly become viable.
We’ve already seen how a single provider’s disruption can ripple across industries. The Change Healthcare and CDK Global incidents exposed the consequences of interconnected systems and supply chains.
Now imagine attackers who no longer need deep familiarity with the technology they’re targeting.
Ransomware transformed cybercrime by introducing a new revenue model: extortion through operational disruption. Attackers then made that model scalable and profitable.
AI could spark the next transformation, not simply by accelerating existing attacks, but by enabling entirely new ways to monetize a compromise.
The biggest threat may not be faster ransomware. It may be a new form of cybercrime we haven’t anticipated, powered by AI and built around a revenue model we haven’t seen before.
– John Dwyer
Identity will become a bigger target, and a bigger liability
In 2027, expect attackers to focus more aggressively on the identity layer, particularly shared identity providers and SaaS platforms.
Organizations have concentrated authentication and access across a relatively small number of providers. That creates efficiency, but also concentration risk.
One successful credential-stuffing campaign, stolen token or provider security lapse could expose hundreds of customers simultaneously.
You can’t patch away concentration risk.
Add AI tools holding long-lived API keys, and the potential blast radius grows.
Boards need to stop treating identity security as a vendor management checkbox. A compromise at a shared provider can quickly become everyone’s problem.
– Neil “Grifter” Wyler
AI will amplify expertise, but it can also erode judgment
AI can make skilled professionals dramatically more productive. The best practitioners use it to accelerate execution without surrendering critical thinking.
The danger comes when people delegate both the work and the judgment.
We’ve all seen the result: the “slop grenade,” low-quality AI-generated output passed along without review, context or accountability.
In cybersecurity, that habit carries far greater consequences than a poorly written report.
As organizations automate more security functions, the risk isn’t simply that AI makes mistakes. It’s that people stop questioning the answers.
AI should accelerate decisions, not replace the judgment behind them.
Human oversight must remain meaningful, with clear accountability for the actions and outputs of AI systems.
– John Hendley
In 2027, security will be measured in minutes, not checklists
Speed will define cybersecurity effectiveness in 2027.
AI is compressing attack timelines. Activities that once took days or hours can increasingly unfold in minutes. Defenders must close the gap between detection and containment.
The question is no longer simply, “Are we secure?”
It’s “How quickly can we detect a compromise, contain it and recover?”
That requires an assume-breach mindset: segmentation, tested backups, identity kill switches and rehearsed incident response.
And not just tabletop exercises that look good in a report. Real preparation means involving the people who will answer the phone and operate the systems at 2 a.m. when an incident hits.
If attackers operate at machine speed, defenders can’t respond at meeting speed.
– Neil “Grifter” Wyler
Defenders need access to the same advances attackers exploit
As AI capabilities accelerate, security leaders must advocate for the tools defenders need to keep pace.
Federal policy and regulation remain uncertain, and organizations cannot rely solely on frontier model providers to determine which capabilities defenders can access.
Attackers will continue finding ways around safeguards. Restricting legitimate defensive research and capabilities risks widening the advantage attackers already have.
We can’t ask defenders to fight AI-enabled threats while limiting their ability to use AI effectively.
Security leaders must push for responsible access to advanced defensive capabilities, including open-weight models, while maintaining appropriate controls and accountability.
– John Hendley
Preparation is the advantage
The cybersecurity challenge in 2027 isn’t simply more AI. It’s the combination of faster attacks, broader exposure and growing dependence on interconnected systems.
Organizations need more than new tools. They need tested defenses, stronger identity controls, faster detection and containment, and leaders willing to invest in the people protecting the business.
The organizations best prepared for 2027 won’t be the ones predicting every attack. They’ll be the ones ready to respond when it happens.






