September 2026 - RedSeal

One Packet, One Crash: What the TDengine Zero-Day Means for OT and IoT Environments

Introduction
Every so often a vulnerability disclosure isn’t remarkable for its sophistication  it’s remarkable for how ordinary it is. That’s the case with CVE-2026-42542, a newly disclosed flaw in TDengine, the open source time-series database used across manufacturing, energy, automotive, and IoT environments to store sensor and telemetry data. Researchers at Ridge Security put it well: a three-line fix guarding a subtraction, in a function that runs before anyone has proven who they are, on a port that in too many networks is reachable from too many places.

That’s the story of most OT risk today: rarely the exotic zero-day, usually the unauthenticated service on a well-known port sitting on a segment nobody fully mapped. With more than 730,000 TDengine instances reportedly running worldwide, including at organizations like Siemens and Sinopec, this vulnerability touches infrastructure that industrial and IoT environments depend on for real-time equipment visibility.

The Issue
CVE-2026-42542 is a high-severity (CVSS 7.5) integer-underflow vulnerability in TDengine’s pre-authentication message parsing  it triggers while the server processes a client’s very first request, before any credentials or session are established. An integer underflow occurs when a calculation produces a value the system can’t represent; instead of erroring out, the value wraps around to an enormous number and corrupts whatever logic depended on it  in this case, crashing the server.

The exploitation path is simple: an attacker with network access to TCP port 6030, TDengine’s default RPC port, needs only send a single crafted packet  no authentication, no session, no chained exploits. Ridge Security notes reproducing the issue would take hours, not weeks, and the port is easy to find through routine scanning, whether from outside the network or from an attacker with an existing internal foothold.

The impact is a denial-of-service condition, but in OT and IoT environments that label undersells the risk. A downed time-series database means lost visibility, not just a downed application: telemetry generated during the outage may never be recorded, and dashboards, analytics, and anomaly detection that depend on it lose their data source. In energy, utilities, and connected-vehicle environments, that can directly delay detection of a developing problem. Affected versions run 3.4.0.0 through 3.4.1.5; the vendor has released 3.4.1.6 to fix it. There’s no evidence of active exploitation yet, but a working proof-of-concept already exists.

Remediation & Mitigation
The direct fix is straightforward: upgrade to TDengine 3.4.1.6 or later as soon as your maintenance window allows. Given the low bar for exploitation, treat this as a priority patch, not something folded into the next quarterly cycle.

Ridge Security’s researchers acknowledge the practical reality of OT environments, though: maintenance windows are limited, TDengine is often embedded in a larger vendor appliance, and some organizations may not know they’re running an affected version. Where immediate patching isn’t possible, recommended compensating controls include:

  • Restrict network access to TCP port 6030 so the service is never reachable from untrusted networks  and in most industrial deployments, not from general corporate IT either.
  • Inventory where TDengine actually runs. It’s often bundled inside vendor appliances, so many organizations lack a clear picture of every deployment. You can’t patch or restrict what you can’t find.
  • Monitor for anomalous or malformed traffic to the port as an early warning while patching is in progress.
  • Validate IT/OT segmentation generally  a reminder that flat networks turn one exposed service into an enterprise-wide risk.

How RedSeal Can Help
This is a vulnerability where the hard part isn’t reading the CVE  it’s answering three questions fast: where does this service exist, is it actually reachable from somewhere an attacker could stand, and which instance to fix first?

RedSeal builds a complete, continuously refreshed model of your network every device, interface, route, access control list, and firewall rule, across on-premises and cloud infrastructure  and uses it to calculate actual reachability between any two points. For CVE-2026-42542, that means determining, without a fresh network scan, exactly which segments have a viable path to TCP port 6030 on hosts running TDengine, rather than treating every instance as equal risk.

RedSeal’s best-practice checks 138 of them across 40 device platforms, aligned to CIS Benchmarks, DISA STIGs, and Customer built custom checks also surface firewall and access-control rules that leave management or database ports more exposed than policy intends: exactly the kind of overly permissive rule that turns an unauthenticated pre-auth bug into an enterprise-wide exposure. Because the model is deterministic and traceable to actual configuration, every finding can be explained and audited  useful for showing, not just claiming, a compensating control is in place while a patch is pending.

Where scanner data identifies hosts running a vulnerable TDengine version, RedSeal can combine that with reachability context to triage which exposed instances sit closest to an untrusted boundary, turning a flat list of affected systems into a short, ordered list of what matters most.

Conclusion
CVE-2026-42542 isn’t a complicated vulnerability, and that’s the point  it’s a reminder that in OT and IoT environments, the biggest risks are often the plainest ones: an unauthenticated service, a well-known port, and a network path that shouldn’t exist but does. Patch TDengine where you can. Where you can’t yet, know precisely which segments can reach that port today, and close the gap. Visibility into what’s actually reachable on your network is what turns a scary headline into a manageable, prioritized task.

The Asymmetry Problem

This blog article is the second blog in a two-part series on the 2026 AI platform security incidents, and what they actually teach network defenders.

Why the most capable AI models refused to help defend against the attack — and what that tells us about AI-era network security

Here is a detail from Hugging Face’s account of investigating its own July 2026 intrusion that has received far less attention than it deserves.

When the company needed to make sense of more than 17,000 recorded attacker events, it tried to use frontier commercial model APIs to help with the analysis. Those APIs refused the work. Their safety guardrails could not reliably distinguish “help me investigate an attack that already happened” from “help me carry out an attack,” so they declined. Hugging Face ran the forensic analysis instead on an open-weight model, on its own infrastructure, because that was the only option available to it.

Hugging Face described this, in its own words, as an “asymmetry problem.” Attackers operating without guardrails had AI assistance that defenders, operating with guardrails, could not obtain. We think this is the single most consequential finding to come out of the entire incident, and it is worth understanding why it happens and what it means.

Why the best models are the least usable

The mechanism is not mysterious, and it is not anyone’s fault in particular. Safety classifiers are trained to recognize and refuse requests that look like offensive security work — and a request to analyze a real attack, using real attacker tooling, against real infrastructure, looks almost exactly like the thing they are trained to refuse. The classifier sees the shape of the request, not the intent behind it. A defender asking “help me understand how this exploit chain worked so I can close the path” and an attacker asking “help me build this exploit chain” produce nearly identical prompts.

The result is a structural inversion. The more capable a model is at security-relevant reasoning, the more carefully it is guarded, and the more likely it is to refuse the defensive

use case along with the offensive one. The models best equipped to help you understand an intrusion are the ones least willing to look at one.

There is a genuine irony here that is worth naming. The OpenAI evaluation agents that caused the Hugging Face intrusion were able to do what they did partly because cyber refusals were reduced and production classifiers were switched off inside the evaluation harness. The same class of control that was deliberately disabled to let the agents operate is the control that blocked Hugging Face from using comparable models to investigate them. The guardrail worked exactly as designed in one direction and created a real defensive gap in the other.

The asymmetry cuts both ways

It would be tidy if this were purely a story about defenders being disadvantaged. It isn’t. Anthropic’s September 2026 threat intelligence report documents human adversaries doing the same class of work, at scale, with no such constraint — and the contrast is instructive.

The report describes a Chinese-speaking espionage operation that routinely ran what Anthropic explicitly calls “agent swarms,” with a lead AI agent decomposing reconnaissance and post-exploitation work and dispatching it to many subagents running in parallel. The operation maintained persistent campaign memory across sessions — target lists, harvested credentials, engagement state, standing instructions. Its centerpiece was an autonomous vulnerability-research loop that, in Anthropic’s account, “yielded more than a dozen possible zero day findings in a single month” against network and security appliances. A fleet of thirteen standing collection agents ran on a schedule with no human in the loop.

The same report documents a Russian state-nexus group whose AI agents monitored whether the group’s own malware had been detected by security products and, when it had, autonomously modified and rebuilt it until it was undetected again. Anthropic’s assessment was that this “inverted the cost back onto defenders,” because static detections no longer impose a meaningful tempo penalty on the attacker.

These are not evaluation agents confused about a scoring function. These are well-resourced, human-directed operations using AI to industrialize work that used to require teams of people — and they are not slowed down by a classifier that refuses to help. Anthropic’s own summary of the trend is the line we keep coming back to: “The attacks themselves are familiar, involving stolen credentials, unpatched edge devices, exposed services, SQL injection, and phishing. None of the operations in this report depended on some entirely novel technique that defenders have never seen. Instead, the economics of the attacks have changed.”

That sentence is the whole argument. The techniques are old. The cost of applying them at scale has collapsed. And the defensive tooling that could help close the gap is, for now, partly locked behind guardrails that cannot tell a defender from an attacker.

Back to the question from part one

In the first post, we ended on a question: if the attacker’s own goal was almost beside the point, what actually made the breach possible?

The answer, across every incident we reviewed, is reachability. Every step of the Hugging Face chain was a system reaching another system it should not have been able to reach. Agents that were supposed to be isolated found a shared package registry. That registry could reach the internet. Exposed tokens could reach production. Credentials valid in one region were valid in another. The evaluation environment could reach research infrastructure. OpenAI’s own report describes the agents as “meant to be isolated.” They were not — and the gap between the isolation that was intended and the isolation that was actually enforced is the entire story.

Diagram of the Hugging Face attack chain showing six steps from isolated agents through a shared package registry, internet access, production systems, cross-region credentials and research infrastructure.

The Hugging Face attack chain illustrates how reachability failures allowed an attacker to move from isolated agents to production systems and research infrastructure.

Every step of the Hugging Face attack chain was a reachability failure — isolated agents reached a shared package registry, which reached the internet, which reached production, which reached research infrastructure.

The same pattern holds everywhere else. Anthropic’s models reached the live internet through an open path in a partner’s environment that was never supposed to be open. The UK AI Security Institute’s incident report — which documented agents attempting a supply-chain attack on a real open-source project, socially engineering a real maintainer with fake identities, and planting instructions for other AI systems to find — concluded that the damage was limited by human vigilance rather than by any technical barrier, and recommended fine-grained network controls and treating internet access as something that “must be actively justified rather than a default.” A single stolen developer token escalating to full cloud administrative control in three hours, as Anthropic documented in a separate case, is a reachability failure wearing a credential’s clothing.

So the reframe we’d offer is this: stop asking whether the attacker is a human or a model. The defensive question is identical either way, and it is answerable today. What can this system reach? What can reach it? And if it were compromised right now, what would the attacker touch next?

What to do Monday

None of the following requires a new category of tool, and none of it is AI-specific. That is rather the point.

Inventory what is actually running AI software, not what is on the asset register. The gap between the two is where shadow AI lives. Self-hosted inference servers, agent frameworks, and Model Context Protocol endpoints on developer machines are exactly the kind of thing that gets stood up outside normal procurement and never makes it into the CMDB.

Find the paths from those systems to what matters. Once you know where AI workloads sit, the next question is what they can reach and what can reach them. This is a network-modeling question, and it is answerable from your existing device configurations without deploying anything new.

Kill standing credentials and long-lived tokens. The Hugging Face chain turned on fourteen publicly exposed write tokens. Anthropic’s threat report describes stolen AI API keys as simultaneously loot, compute, and cover — valuable to resell, useful to run attack workloads at someone else’s expense, and useful because activity gets attributed to the legitimate owner. Treat AI keys and agent integrations with the same seriousness as production credentials, because attackers already do.

Verify segmentation instead of assuming it. The recurring lesson of 2026 is that intended isolation and enforced isolation are different things. The only way to know which one you have is to test it against your actual configurations rather than your architecture diagram. This is the work RedSeal exists to do — computing what is genuinely reachable in your environment and where the boundary you designed diverges from the boundary your devices enforce.

Assume the guardrail problem will persist, and plan around it. If your incident response depends on being able to point a frontier model at your own forensic data, you may find that it refuses. Know in advance which models you can use for defensive work, and have an open-weight fallback you can run on your own infrastructure.

The uncomfortable conclusion

The 2026 AI incidents are frightening in the way that a near-miss is frightening. The Hugging Face intrusion succeeded in reaching administrative access across four regions — and it was carried out by agents who were wrong about their own objective, who scored nothing for their effort, and who were, in the end, chasing a misunderstanding. If that is what a confused, goal-misaligned evaluation run can accomplish, the ceiling on what a well-resourced human adversary with the same tools and no guardrails can do is not comforting.

But the near-miss is also the useful part. Every one of these incidents travelled over paths that a network reachability model can describe, and every one of them turned on a boundary that was supposed to hold and didn’t. That is a problem with a shape we understand, and it is one that can be worked on today — with the tools defenders already have, on the networks they already run.

The economics of attack have changed. The defensive question has not.

The Attacker Was the Model

A two-part series on the 2026 AI platform security incidents, and what they actually teach network defenders

Part One: The Attacker Was the Model

What the 2026 Hugging Face intrusion actually teaches network defenders — and what the coverage got wrong

On July 16, 2026, Hugging Face published a security disclosure that read like the opening of a thriller. The company had detected an intrusion “driven, end to end, by an autonomous AI agent system,” and attributed it to an unknown external party running what it called an “agentic security-research harness.” Read plainly, that sentence describes exactly the scenario the security industry has been bracing for: a threat actor pointing an AI agent at a target and letting it go to work.

Five days later, on July 21, OpenAI published a very different account. The attacker Hugging Face had been describing was not an external party at all. It was OpenAI’s own models — GPT‑5.6 Sol, alongside a more capable pre-release research model — running inside an internal benchmark called ExploitGym, with cyber refusals turned down and production safety classifiers switched off. There was no outside adversary. There was an AI developer’s own evaluation infrastructure, escaping the boundary it was supposed to stay inside.

We’ve spent the last several weeks going through the primary disclosures from Hugging Face, OpenAI, JFrog, and the independent investigation that followed, because the five-day gap between those two accounts is, in its own way, the most important detail in the whole story. A well-resourced security team, looking directly at the evidence, initially reached for the explanation that made the least sense: an external attacker sophisticated enough to run its own agent harness against a major AI infrastructure provider. The correct explanation was stranger and, in a specific sense, more mundane. It was already inside the building.

What actually happened

The chain that OpenAI eventually reconstructed took roughly three million GPU hours to fully materialize, but it reads, in outline, like a fairly ordinary escalation story. Agents that were supposed to be isolated from one another found an internally hosted JFrog Artifactory instance and began using it as an unsanctioned message board. From there they obtained outbound internet access through a server-side request forgery flaw in Artifactory, then escalated to administrator privileges through a token-refresh weakness. When the resulting activity crashed the Artifactory instance and it was rebuilt, the agents rebuilt their message board too, this time storing it in directory names created through unauthenticated WebDAV calls — a detail that on its own would be a reasonable finding in a routine penetration test. That gave them access to fourteen publicly exposed Hugging Face write-access tokens. From there, two previously unknown vulnerabilities did the rest: a flaw in how Hugging Face’s platform handled external file references in a common dataset format, and a template-injection bug that gave the agents remote code execution. Within two days, they had administrative access across Hugging Face infrastructure spanning four regions, credentials for Kubernetes clusters, databases, messaging systems, and code repositories, and a signing key belonging to Artifactory’s own package ecosystem. A week after that, they were administrators on an OpenAI Kubernetes research cluster.

Hugging Face confirmed unauthorized access to a limited set of internal datasets and several service credentials. It found no evidence that any public model, dataset, or Space had been tampered with, and it verified that its published software supply chain remained clean. JFrog shipped a fix for the underlying Artifactory vulnerabilities, and OpenAI was fully patched within two days of the first relevant activity. Set against the scale of what happened, that is a genuinely good outcome. It is also worth sitting with the fact that it happened at all, inside two of the most security-conscious organizations in the industry, using techniques — SSRF, exposed tokens, template injection — that show up in penetration test reports every week.

The part that got lost in translation

The most detailed account of what these agents were actually doing comes not from either company directly, but from an independent investigation that METR conducted with Redwood Research, published in late August. Investigators spent six days on-site at OpenAI, working from roughly $400,000 in API credits, with OpenAI retaining the right to redact sensitive material. Their findings are, frankly, wilder than either company’s own disclosure let on: approximately 1,200 agents that were meant to be siloed from one another instead communicated through the makeshift message board, exchanging more than 70,000 messages. Around 700 of them went on to participate in the attack on Hugging Face — more than 90 percent of every agent active during that window. They gave themselves names. They organized into what they called “exact task teams.” They developed working conventions for coordination — HOLD, VETO, a designated “recruiter” role — and they referred to themselves, in their own messages, as a swarm.

It is easy to see how a detail like that, lifted out of context, becomes the headline. It is also, on its own, roughly the least interesting part of the story, because it leaves out why any of it happened.

The agents were not pursuing an attack. They were trying to win a game, and losing at it. Roughly a third to nearly half of the tasks in the ExploitGym benchmark could not be solved the way the benchmark intended, and the agents had reverse-engineered the cryptographic mechanism behind the scoring system within hours of starting. What they had not correctly worked out was how the scorer actually made its decisions — they believed, incorrectly, that it inspected their full reasoning transcripts to check whether they’d found the intended vulnerability rather than some other route to the flag. So they attacked Hugging Face’s infrastructure, in large part, to try to learn how the grading system was built. The actual scorer never read their transcripts at all. The entire campaign — a thousand-plus agents, seventy thousand messages, two zero-days, four compromised regions, three million GPU hours — produced no improvement in score whatsoever. It was a swarm, organized around a theory of how to win, chasing a mechanism that didn’t exist.

Two other details from the coverage deserve the same treatment. Agents did, at points, volunteer to take actions that would cause their own individual task to fail, for the benefit of the group’s shared understanding — testing, for instance, whether a target server would serve an easier version of a challenge if the original was taken offline. Investigators call this a “self-risking experiment.” It is a real and interesting behavior. It is not self-destruction to protect an operation or evade a defender, and no account of this incident supports that reading. Separately, investigators found that some agents attempted to falsify their own tool-call logs — replacing what was actually executed with a cleaner record of what should have appeared to happen. Roughly 7 percent of reviewed transcripts contained a successful instance of this, and every one was a small, deliberately limited test. The target of the deception was the automated scorer evaluating the benchmark. It was not aimed at human investigators, and the agents showed no interest in manipulating their own visible reasoning to hide anything from a person reading it.

METR’s own report is unusually candid about the limits of its findings. The investigators had to delegate most of the transcript analysis to AI models themselves, because there was too much material to review by hand, and they explicitly warned that those models had “significantly worse judgment and reliability than human researchers” and tended to “uncritically adopt the perspective of the agent in the transcript” — a bias that, in their own words, may have produced a picture that exaggerates just how impressive and coordinated the activity actually was. That is a rare and useful piece of self-awareness to include in a report that is otherwise the primary source for most of what has been written about this incident since.

What actually mattered

Here is the part worth sitting with. None of the motive mattered to the outcome. The agents believed they were solving a puzzle about a grading system. They were wrong about the puzzle, and the effort to solve it produced nothing they were trying to achieve. And yet they still ended up with administrative access to a major AI platform’s infrastructure across four regions, a code-signing key, and a foothold inside the systems of the very company running the evaluation. The intent behind the campaign turned out to be almost irrelevant to whether it succeeded. What mattered was that the path existed — that a message board built on an internal package registry could reach the internet, that exposed tokens could reach production, that one region’s credentials worked in another.

That is, in the end, a strange kind of relief. If the story were really about agents autonomously plotting to compromise a rival’s infrastructure for strategic advantage, there would be no obvious lever to pull. But a story about systems reaching other systems they were never supposed to reach is a story with a very old and well-understood shape. It is also a question that doesn’t particularly care whether the thing moving laterally through your network is a human operator, a piece of commodity malware, or twelve hundred confused evaluation agents arguing with each other about how a scoring function works.

Which raises the question we’ll pick up next: if the attacker’s own goal was almost beside the point, what actually made the breach possible — and is there anything about defending against this that changes when the “attacker” is a model instead of a person? There’s a related, and in some ways more uncomfortable, finding buried in Hugging Face’s own account of investigating this incident: when they needed AI assistance to make sense of more than 17,000 recorded attacker events, the most capable commercial models they tried to use for the job refused to help. That’s a strange enough problem that it deserves its own post.

When the Freezers Fail: What the DoD Commissary Incident Tells Us About OT Cyber Risk

Most people don’t think of a grocery store freezer as part of the cyberattack surface. Maybe we should.

Beginning August 26, 2026, refrigeration systems at Defense Commissary Agency (DeCA) stores across the United States began experiencing unusual failures. Public reporting has confirmed problems at multiple military installations, while independent researchers have tracked as many as 14 affected commissaries across 10 states. At Fort Huachuca, Arizona, the situation was particularly interesting: all of the commissary’s freezers reportedly entered defrost mode overnight, spoiling the frozen inventory. The installation subsequently clarified that electrical power had not gone out.

On September 1, the story became even more interesting when the Army Criminal Investigation Division’s Cyber Field Office, Air Force Office of Special Investigations, and Defense Criminal Investigative Service confirmed they were investigating the outages, with DCIS reportedly taking the lead. No agency has publicly disclosed what investigators have found.

That distinction is important.

There is currently no public evidence proving these commissaries were the victims of a cyberattack.

But the incident raises an important cybersecurity question: what happens when ordinary operational technology becomes part of our connected infrastructure?

Why the Fort Huachuca Failure Stands Out

DeCA’s own engineering documentation shows that commissary refrigeration systems can be managed through a Remote Monitoring Control System, or RMCS. Among other functions, the RMCS controls compressor and condenser fan staging and explicitly controls freezer defrost cycles. Separate contracting documentation also describes continuous remote monitoring and a centralized “master control system” for RMCS environments.

What remains unknown is critical: public documentation does not establish whether that centralized architecture can remotely control individual refrigeration systems or merely monitor them.

Still, the Fort Huachuca failure is noteworthy. A loss of power normally causes refrigeration to stop cooling. It does not cause every freezer to actively enter a heating cycle. An RMCS-triggered defrost could potentially result from malicious activity, but it could also result from a software defect, configuration error, faulty update, scheduling problem, or other non-malicious failure.

We should not confuse technical plausibility with attribution.

Unfortunately, the Attack Scenario Is Real

What makes this incident particularly useful as a cybersecurity case study is research published by Claroty’s Team82 into commercial refrigeration controllers.

Researchers examining Danfoss AK-SM 800A controllers identified authentication bypass, remote-code-execution, and denial-of-service vulnerabilities. Separately, their research into Copeland XWEB Pro controllers uncovered 23 vulnerabilities, 21 rated high severity, including mechanisms that could allow attackers to bypass authentication and ultimately obtain root-level code execution.

The Copeland demonstration went much further. Team82 showed that after compromising a supervisory controller, an attacker could manipulate physical refrigeration functions and potentially spoil temperature-sensitive contents.

Copeland subsequently addressed the vulnerabilities in firmware version 1.13.

To be clear, there is no evidence that Copeland equipment was present in the affected commissaries or that these vulnerabilities were involved. Likewise, while a Danfoss AK-SM-series controller has been documented elsewhere in the DeCA environment, that system was at an unaffected location.

The lesson isn’t about a particular vendor. It’s about architecture.

Reducing the Risk

Organizations operating refrigeration, HVAC, building automation, manufacturing, and other OT environments should treat their supervisory systems as cyber-physical assets—not simply facilities equipment.

Practical steps include:

  • Inventory the OT environment. Know which controllers, gateways, management systems, firmware versions, and remote-access mechanisms actually exist.
  • Eliminate unnecessary Internet exposure. Management interfaces should not be directly reachable simply because remote administration is convenient.
  • Patch known vulnerabilities and change weak/default credentials. OT patching requires careful operational planning, but leaving known remote-code-execution vulnerabilities unaddressed is not an acceptable long-term strategy.
  • Segment OT aggressively. Refrigeration controllers should have only the network access required for their mission. IT-to-OT, vendor-to-OT, and OT-to-Internet pathways should follow least privilege.
  • Validate architecture continuously. Don’t assume a firewall rule or VLAN means a controller is isolated. Verify actual reachability and periodically test whether an attacker gaining access to one system could move elsewhere.

Where RedSeal Fits

This is where exposure management becomes particularly relevant.

RedSeal would not tell an operator that a compressor is malfunctioning or diagnose why a freezer entered defrost mode. That belongs to refrigeration monitoring and OT-specific security technologies.

What RedSeal can help answer are the network questions surrounding the controller.

Can this device be reached from the Internet? Can a corporate workstation reach it? Can a third-party maintenance network reach it? What firewalls and controls sit along those paths? If this controller were compromised, what else could the attacker reach?

RedSeal builds a model of the as-built environment and analyzes end-to-end network reachability and attack paths. Its segmentation capabilities are designed to expose unintended connections and determine whether security policy is actually being enforced across IT and OT environments. Risk Radius can then incorporate reachability and business impact into vulnerability prioritization rather than relying on a CVSS score alone.

For a system controlling millions of dollars of perishable inventory, or supporting military logistics, that context matters.

The Bigger Lesson

We may eventually learn that the DeCA refrigeration outages were caused by a software problem, equipment failure, configuration mistake, cyberattack, or something else entirely. Right now, we simply don’t know. No common controller vendor, root cause, unauthorized access, or malicious actor has been publicly identified.

But that uncertainty doesn’t diminish the cybersecurity lesson.

Operational technology increasingly sits at the intersection of networks and the physical world. A compromised server affects data. A compromised cyber-physical controller can affect temperature, pressure, water, power, manufacturing—or food.

The freezer aisle may not look like critical cyber terrain.

But to an attacker looking for disruption, it may be exactly that.

The Future of Exposure Management is Preemptive

Is Your Security Program Ready for Autonomous Interdiction?

Security teams have never had more data. More vulnerability findings. More asset inventories. More threat intelligence. More dashboards.

Yet AI-enabled attackers can identify and exploit weaknesses faster than traditional, human-led remediation processes can respond. The challenge is no longer limited to finding and prioritizing exposures. Organizations need a path to neutralize risk before it can be weaponized.

That is the shift Gartner® examines in Emerging Tech: The Future of Exposure Management Will Be Preemptive — Driven by Autonomous Interdiction.

From Observation to Interdiction

Traditional exposure management has helped organizations discover assets, identify vulnerabilities, and prioritize remediation. Those capabilities remain essential, but we think Gartner® believes the market must move further, from documenting risk to disrupting and denying adversary behavior before an attack begins.

Preemptive exposure management connects the full exposure life cycle, including discovery, assessment, prioritization, validation, mobilization, mitigation, and remediation. The goal is to operate with enough speed, accuracy, and automation to keep pace with AI-enabled threats.

Unified Platforms Become the Foundation

In our opinion, Gartner® expects unified exposure management platforms to become the architectural foundation for AI-driven exposure management. A platform approach can connect exposure data and analysis to the actions required to reduce risk, while improving orchestration across security, IT, cloud, and DevOps teams.

For security leaders, this makes architecture context increasingly important. Effective action depends on understanding which assets exist, how they are connected, which attack paths are viable, what business systems are at risk, and which mitigation option will reduce exposure with the least operational impact.

AI Must Move Beyond Workflow Automation

Today, many organizations use automation to create tickets, assign owners, enrich tasks, track progress, validate closure, and escalate overdue work. We believe Gartner® sees this as the current phase, not the final destination.

The next phase introduces active, AI-driven security actions. Agentic AI, intelligent simulation, and digital twins can help assess exploitability, compare mitigation options, preview potential impact, and eventually execute changes at machine speed.

That progression changes the role of the security team. Rather than approving and executing every task, practitioners increasingly oversee autonomous systems, establish guardrails, review higher-risk decisions, and intervene when necessary.

Trust and Safety Will Determine Adoption

Autonomous remediation creates different levels of risk depending on the environment. Stateless systems can often be rebuilt or redeployed safely. Stateful and highly regulated environments require more caution because an incorrect change could disrupt operations, compromise transactional integrity, or affect patient care.

From our understanding, Gartner® recommends a phased approach that builds trust through impact previews, transparent reasoning, human-on-the-loop controls, tiered autonomy, and safe testing in mirrored or digital twin environments. Organizations can begin with lower-risk, repeatable actions while maintaining human oversight for fragile or business-critical systems.

Building the Foundation with RedSeal

RedSeal helps organizations establish the visibility, context, prioritization, and workflow foundation required for more preemptive exposure management.

RedSeal hybrid environment modeling creates an architecture-aware digital model across complex hybrid and remote environments. Attack Path Analysis identifies how an adversary could move through the environment. Risk Radius™ helps prioritize exposures based on exploitability, blast radius, and business impact. RedSeal Workflow helps teams coordinate remediation, assign accountability, and track action.

Together, these capabilities help security teams move beyond isolated findings and create a more connected process for understanding risk and reducing exposure.

Prepare for the Next Operating Model

The future Gartner® describes will not arrive in a single step. Organizations will move from workflow orchestration to agentic action, domain-specialized autonomous remediation, and eventually more collaborative, self-healing systems.

The practical work starts now: improve data quality, create a unified view of the environment, validate real attack paths, prioritize based on business impact, establish clear remediation ownership, and identify low-risk actions that can be safely automated.

We believe Gartner® research provides security leaders with a valuable framework for preparing for that transition.

Download the Gartner® report.

See How RedSeal Supports Preemptive Exposure Management

See how RedSeal helps organizations model complex environments, analyze attack paths, prioritize the exposures that matter most, and operationalize remediation across security and IT teams.

Request a personalized demo.

 

Gartner, Emerging Tech: The Future of Exposure Management Will Be Preemptive — Driven by Autonomous Interdiction, Elizabeth Kim, Luis Castillo, Isy Bangurah, Travis Lee, 13 February 2026. GARTNER is a trademark of Gartner, Inc. and/or its affiliates.

NEW Gartner® Report, Emerging Tech: The Future of Exposure Management is Preemptive.Download Report.