Speed gap: defense lags behind
„The number of vulnerabilities and the speed of exploitation have exceeded the industry's processing and patching capacity; the cybersecurity institutional system cannot keep up with the pace.”
Counter-narrative
The proportion of truly exploited vulnerabilities is small and manageable with risk-based prioritization. This is represented by the NIST prioritization logic and the KEV-based approach: CVEs on the KEV list and those affecting critical software are prioritized, while others fall into the 'Not Scheduled' category.
Voices
Europol (IOCTA), NIST, CISA, Five Eyes agencies, brit NCSC, professional patch-analysis portals
Analysis
This is the profession's self-image: defenders are no longer fighting attackers but their own processing capacity. It is fed by record patch bundles, the NVD's pullback and three-day CISA deadlines. Agencies have an interest in it as an argument for more resources, and vendors as an argument for automation products. Its weak point is that the CVE count and real risk are conflated. In the monthly bundles, the number of already exploited zero-days ranges between one and three, so the narrative easily produces panicked, unprioritized reactions.
What's at stake: The stakes are whether defense can shift from full coverage to risk-based operations before automated exploit production pulls permanently ahead.
Narrative sub-threads
The CVE infrastructure cannot keep up
Vulnerability-registry bodies are being forced to prioritize, narrow their services and restructure under the mass of submissions. NIST is also weighing automation. CVE submissions have risen 263 percent since 2020, and the registries are further burdened by AI-generated false reports.
The topic is institutional adaptation by NIST, NVD, the CVE program, CISA, and ENISA, not attacker speed or patch sizes.
↗ rising · evidence: strongExploitation window shrinking to hours
In multiple documented cases, only hours or a few days passed between disclosure or patching and exploitation. This occurred with Marimo within ten hours, BeyondTrust within 24 hours, and SmarterMail two days after the patch. Authorities and researchers are responding with stricter deadlines. CISA mandates a three-day patch deadline, CERT-In proposes a 12-hour obligation, and Anthropic and Google Project Zero are shortening the disclosure deadline to 30 days.
Document the timing of exploitation of specific vulnerabilities and deadline regulations, not the number of vulnerabilities or the general attack capabilities of AI agents.
▲ at peak · evidence: strongRecord-sized patch releases and KEV growth
Microsoft's monthly patch packages increased from several hundred to over a thousand in the second half of 2026. July saw 570–622, September 973–974, October a record 1,413–1,415, and November 1,094 vulnerabilities closed. The KEV catalog of actually exploited vulnerabilities expanded by approximately 20 percent in 2025, exceeding 1,480 entries. In the third quarter of 2026, it received 100 new entries, 33.3 percent more than in the second quarter.
Based on quantitative evidence: Microsoft Patch Tuesday records, KEV statistics, and the DBIR, not institutional reaction.
↗ rising · evidence: moderatepace of AI-agent attackers, official alert
According to authorities and threat hunters, AI agents shorten certain phases of attacks to hours. The FBI states that privileged access can be obtained within three hours using ARTEX, while Google GTIG observed credential theft in less than six hours. Europol states that AI-supported crime opens a speed gap that law enforcement cannot close. The NCSC considers AI-based ransomware attacks on critical infrastructure extremely likely by 2028.
The acceleration of the entire attack process is based on official and threat intelligence assessments and forecasts, not on the exploitation timelines of individual CVEs.
Sub-thread stories
The CVE infrastructure cannot keep up
Exploitation window shrinking to hours
Record-sized patch releases and KEV growth
pace of AI-agent attackers, official alert
Who is affected and how
- security leader
- Prioritize based on the KEV list and exposed edge devices instead of aiming for full coverage. Measure the repair time of exploited bugs, not all bugs.
- security operations
- Due to exploitation within 24 hours, critical internet-accessible systems require an accelerated patching channel and pre-written scenarios for mitigating measures.
- developer
- Exploits built from patches with AI agents shorten the time between a patch's release and exploitation. Therefore, internal vulnerability management and patch release must be automated.
- public-sector decision-maker
- With the NVD narrowing its scope, it is a risk if domestic organizations rely on a US data source. It is worth supporting EU or in-house enrichment capabilities.
Hungary
During the examined period, the narrative has little or no domestic footprint: the list contains no [HU] story, and no independent Hungarian metric on patching capacity is included.
Weekly trend
Turning points
Supporting stories
Machine analysis based on story titles and summaries, with a verification pass. The source of the facts is the article linked at each story.
What do you make of this piece?
The split shows after 20 votes.
Why? Your answer is not shown publicly; it helps the editors.