The gap between discovering software vulnerabilities and fixing them is growing, increasing organizations’ risk of losses from cyberattacks, according to a report released Tuesday by Black Duck, an application security company in Burlington, Mass.
The report, based on an analysis of 65.5 million application security testing findings across roughly 13,100 applications at 533 organizations, found an alarming trend between June 2025 and May 2026.
In June 2025, organizations were recording 11 new findings for every one they closed. By May 2026, the ratio was 18:1, though it peaked at 22:1 in December 2025.
Organizations once counted on having months to patch a disclosed flaw, the Readiness Report for 2027 explained. Among vulnerabilities tracked by the Zero Day Clock, the average time from disclosure to first known exploitation has fallen below a day, the report noted. AI models can also produce working exploits with less effort than before.
“Discovery is scaling faster than remediation,” explained Black Duck Senior Director Shandra Gemmiti. “Code volume, AI-assisted development, wider test coverage, and AI-powered vulnerability discovery are all adding findings, but most remediation still runs on human-paced, manual workflows.”
“The exploit gap is about speed,” she told TechNewsWorld. “In 2018, a vulnerability was typically first exploited 2.6 years after it was disclosed. Today that takes less than a day, and frontier AI models are pushing it toward minutes.”
“Meanwhile,” she continued, “the fastest-fixed findings still take about a month at the median to close. Attackers are on an hourly clock, and defenders are on a monthly one.”
“This isn’t a staffing problem,” she added. “No amount of headcount can keep up with backlog growth measured in millions of findings a month.”
Why Patching Takes Time
However, Erich Kron, a CISO advisor at KnowBe4, a digital workforce cybersecurity company in Clearwater, Fla., argued staffing can be an issue when mitigating vulnerabilities.
“For each vulnerability, time must be spent testing, prioritizing, and installing patches or mitigations,” he told TechNewsWorld. “Many organizations do not have the staffing to absorb the extra workload needed to deal with vulnerabilities while continuing to perform the other necessary duties.”
“There is a limited number of hours in a day that many organizations can allow downtime, or even risk potential downtime, so [fixes] must be done carefully, and that takes time, often creating a backlog,” he explained.
Ryan McCurdy, vice president of marketing at Liquibase, a database-change automation company in Austin, Texas, pointed out that security teams can identify a problem in minutes, but engineering teams still have to determine whether they’re exposed, understand the dependencies, test the fix, and deploy it without breaking production. “That becomes particularly difficult when you’re dealing with critical systems,” he told TechNewsWorld.
AI is accelerating this problem from both directions, he added. “Attackers can find vulnerabilities faster, while developers are generating more code and introducing more dependencies,” he said.
“I don’t think the answer is simply telling organizations to patch everything faster,” he continued. “They need to prioritize based on actual exposure and risk, then make the process of getting fixes safely into production much more efficient.”
Fixing What’s Important
Because organizations are deluged with vulnerabilities, the report noted that prioritizing what to fix has become a central issue for security teams.
“About 90% of findings are never fixed, so every team is already deciding what gets fixed and what waits. The question is whether they decide deliberately,” said Black Duck’s Gemmiti.
One way organizations decide what gets fixed is by severity scores, which are standardized ratings of how serious a security flaw is.
But Gemmiti argued that prioritization shouldn’t rest on severity scores alone because exploitability doesn’t follow severity. She cited Black Duck Security Advisories showing known exploits existed for 45% of high-severity, 39% of medium-severity, and 31% of low-severity vulnerabilities.
“Teams must adapt to understanding as much as possible about vulnerabilities,” added Brian Wilson, CISO of SAS, an analytics and artificial intelligence software company in Cary, N.C.
“Traditional Common Vulnerability Scoring System [CVSS] scores can be supplemented with additional data points,” he told TechNewsWorld, “such as the Exploit Prediction Scoring System [EPSS], which estimates the likelihood of exploitation, and CISA’s Known Exploited Vulnerabilities [KEV] catalog, which identifies vulnerabilities already being exploited.”
Prioritization is becoming a critical part of vulnerability patch management, observed Andi Ursry, a threat intelligence analyst with Blackpoint Cyber, a managed detection and response company headquartered in Denver.
“A severity score tells an organization how serious the vulnerability can be,” she told TechNewsWorld. “Prioritization determines how serious it is for the organization and how quickly they need to act.”
“A high-severity vulnerability being actively exploited against an internet-facing remote access tool likely warrants immediate action ahead of a critical-severity vulnerability not actively exploited in a component that is disabled by default,” she explained.
End of an Era
However, Larry Maccherone, founder and CTO of Lumenize, a builder of agentic software-engineering and DevSecOps platforms in Raleigh, N.C., discounted the value of prioritizing vulnerabilities.
He cited a Carnegie Mellon study that found that unless a vulnerability was rated a 9 or 10, the correlation of risk to severity score was no better than random chance. “So our ability to figure out what’s likely to be exploited is really pretty bad,” he told TechNewsWorld. “And it always has been.”
Furthermore, he maintained that any effort to prioritize vulnerabilities isn’t worth it. While head of application security at Comcast, he found he could fix 10 vulnerabilities in the same time it took to triage one.
“So it’s more expensive by an order of magnitude to triage vulnerabilities than to fix them,” he said. “And fixing 10 of them is going to at least lower your probability of being attacked to one-third what it was before triage.”
Dwayne McDaniel, a developer advocate at GitGuardian, a non-human identity security company, asserted that the Black Duck report is another wave of evidence that the era of reactive security is over.
“We’re watching decades of technical debt come due,” he told TechNewsWorld. “We have been deploying infrastructure and applications for years with the assumption that we’d always have time to patch things later, as needed. Now, attacks that once took weeks of human planning unfold in minutes, sometimes compromising critical systems in seconds, and no human-driven response process can keep pace.”
“We’ve known for years that credentials are sprawled across our systems, hardcoded into applications, and granted permissions far beyond what they need,” he continued. “That remains the lowest-hanging fruit and the most common way these attacks succeed. None of this is really new aside from the speed at which attackers can exploit the architectural weaknesses we’ve allowed to accumulate.”
“We need to stop treating every incident as another individual fire to put out and accelerate the move, right now, toward zero trust architectures that limit access by design,” he said. “The time we thought we had to fix these problems has run out.”



