Rapid7 Finds Attackers Exploit Select Vulnerabilities Within 72 Hours

| 2 Min Read
Rapid7 tracked 43 newly exploited vulnerabilities in Q3 2026. Most attacks targeted remotely reachable flaws that needed no login or user action.

A growing pool, a selective threat

Attackers aren’t exploiting every newly disclosed flaw. Rapid7’s Q3 2026 findings point to a narrower risk for defenders: adversaries are selecting vulnerabilities that are easy to reach and quick to weaponize, often within about 72 hours of disclosure. The distinction matters because the volume of flaws demanding attention can make every item seem equally urgent, even when attackers aren’t treating them that way.

The company tracked 43 newly exploited vulnerabilities during the quarter, a 19% increase from a year earlier. That rise understates how quickly the pool of potential targets is expanding: high-severity CVEs, rated 7 to 10, grew 133% year over year, while public proof-of-concept code increased about 86%. Those figures describe different parts of the problem: more serious flaws are being disclosed, and more code that could help test or exploit them is becoming public.

“The pile of things that could be exploited is growing far faster than the set of things attackers are actually choosing to exploit,” said Douglas McKee, Rapid7’s director of vulnerability intelligence.

That gap is useful, but it shouldn’t be mistaken for reassurance. A selective attacker can still create serious risk by concentrating on flaws that are accessible and practical to exploit. For defenders, the challenge is sorting the expanding list of known weaknesses according to how an attacker could reach them in a specific environment.

Disclosure can quickly turn into exploitation

Rapid7 found that 71% of the exploitation it tracked began within a week. Most underground forum posts referencing CVEs appeared within days of disclosure, after the flaws had already been covered by vendor advisories or added to the US Cybersecurity and Infrastructure Security Agency’s Known Exploited Vulnerabilities catalog. The pattern suggests rapid repackaging of known vulnerabilities, rather than a surge in sales of previously unknown zero-days. Public disclosure can put defenders on notice, while also giving attackers a starting point for assessing a flaw and adapting available information.

McKee said attackers can move from disclosure to real-world use in roughly three days. AI tools may help surface flaws faster than security teams can review them, he said, but adversaries remain selective about which vulnerabilities they weaponize. The short interval puts pressure on the handoff between learning that a flaw exists and deciding whether an exposed system needs immediate attention.

The most attractive targets share a clear profile: they’re reachable over a network, require no authentication and need little or no action from a victim. Nearly 80% of the vulnerabilities exploited during the quarter fit that description, Rapid7 reported. Each characteristic reduces friction for an attacker: a network reachable service can be approached remotely, and the absence of login requirements or user action removes steps that might otherwise block an attempt.

This is more significant than it looks. A vulnerability’s severity rating describes one aspect of risk; exposure and the conditions needed to exploit it help determine whether that risk is immediate for a particular organization. (And this is the part most people overlook.) A high score alone doesn’t tell a team whether an affected system is reachable from outside, or whether an attacker would need a user to make a mistake.

Exposure should shape patch priorities

For security teams, the practical response isn’t simply to speed up every patch cycle. It’s to identify systems exposed to the attack profile adversaries favor, then decide how long each system can remain vulnerable. That means assessing exposure in the context of the organization’s own environment and setting a tolerable number of days before remediation. A patch deadline based on that assessment can account for both the vulnerability and the system’s accessibility.

“If your patching cadence still has the word ‘monthly’ in it, you’re defending against an adversary who abandoned that schedule a long time ago,” McKee said.

That’s a pointed warning, though it doesn’t mean every organization can patch everything immediately. Teams often have to weigh urgency against the time and coordination needed to change production systems. The findings make a case for sorting work by reachable exposure and exploit conditions, so that the most accessible risks don’t wait behind less urgent updates in a routine queue.

If you’re working in this space, the practical question is whether your asset inventory and patch process can identify those exposed systems quickly enough. A faster schedule helps only if teams know which systems are affected and can judge the consequences of leaving them exposed. The operational gap, if there is one, may sit between vulnerability intelligence and the people responsible for remediation.

AI draws concern, but human choices remain

AI-enabled attacks ranked as the top threat concern in CSO parent company Foundry’s forthcoming Security Priorities report, with 45% of CISOs rating their concern high. Rapid7, however, found AI involved operationally in about 8% of the attacks it tracked. The contrast suggests that concern about AI’s potential is running ahead of its observed role in this particular set of attacks.

In those cases, McKee said, AI handled tactical steps such as selecting the next technique or deciding whether a target warranted more effort. A human still chose whom to pursue and why. That division matters: automation can speed up parts of an operation without removing the human judgment behind target selection.

UnderDefense CEO Nazar Tymoshyk expects AI’s role to grow. He said Russian advanced persistent threat groups already use AI-powered attacks and may switch among models, including China’s DeepSeek, which he described as having few constraints or guardrails. Agentic tools can also give attackers access to a broader range of technical expertise, from database administration to CI/CD pipelines, than a conventional team may have on hand. Those are expectations about how attacker capabilities could develop, rather than evidence that AI already drives most exploitation in Rapid7’s data.

More automation could change how quickly attackers assess targets or choose a next step. Yet the quarter’s findings point to a durable operational signal: prioritize systems that are remotely exposed and require no login or victim interaction. Attackers’ speed matters. So does their preference for a specific vulnerability profile, which gives defenders a sharper basis for setting patch deadlines and deciding where urgency is highest.

Not every disclosure will become an attack. But when exploitation can begin within days, teams need a clear way to identify which known flaws are exposed in their own environments and act on that assessment. The advantage here is practical, not dramatic: a more informed deadline for the systems attackers are most likely to reach.

Source: John Rodriguez · www.csoonline.com

Comments

Please sign in to comment.