Google Threat Intelligence Group (GTIG), as summarized by SecurityWeek, reported that monthly vulnerability disclosures rose from 5,045 in January 2026 to 10,740 in August. It also reported more high-risk disclosures and a higher rate of observed exploitation. Those figures do not mean that every new CVE is dangerous: automated CVE assignment can inflate totals, and GTIG said just 0.23% of 2026 disclosures had been observed exploited. The reported shift is in both the pace of discovery and the kinds of flaws being found, with important limits on what the available evidence can prove.
What changed in vulnerability disclosures?
SecurityWeek’s September 30, 2026 account of GTIG analysis covers disclosures from January 2025 through August 2026. For 2026, GTIG reported a rise in monthly disclosures from 5,045 in January to 10,740 in August. Its count of high-risk disclosures rose from 131 in January to 350 in August, a reported increase of 167%.
As an Amazon Associate I earn from qualifying purchases.
GTIG’s high-risk ratings are its own categorization, not CVSS scores. The figures are reported through SecurityWeek; the original GTIG report and its methods were not independently reviewed here. The increase is therefore best read as a change in the counts GTIG tracked, not as a directly verified measure of the number of newly dangerous flaws.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →SecurityWeek’s report quotes GTIG saying: “We found that AI is measurably changing not just the pace of vulnerability discovery and exploitation, but also the types and typical risk profiles of vulnerabilities that are being discovered.”
#1 Best Overall
Do higher CVE counts mean more danger?
No. A disclosure count measures reported or assigned vulnerabilities; it does not show how many are exploitable in real-world conditions or being used by attackers. SecurityWeek’s account notes that automated CVE assignment in open-source ecosystems can increase raw totals. It cites roughly 5,000 Linux-kernel CVEs from January through August 2026, while reporting no in-the-wild zero-day exploitation observed for that set.
GTIG reported that 0.23% of vulnerabilities disclosed in 2026 had been observed exploited. That small share is a useful counterweight to the rising totals, but it does not mean all unobserved vulnerabilities are harmless, or that every flaw carries the same risk. Exposure, exploitability, affected systems, and evidence of exploitation matter more to an organization’s immediate decisions than the headline CVE count alone.
What does the exploitation data show?
GTIG reported 141 distinct exploited vulnerabilities from January through August 2026, compared with 127 across all of 2025. The reported monthly averages were 18 and 10.5, respectively. Because the 2026 period covers eight months and the 2025 period covers a full year, the distinct-vulnerability totals are not like-for-like annual counts; the monthly averages provide additional context.
The report also distinguishes zero-day exploitation—attacks before a vulnerability is publicly disclosed—from n-day exploitation, which uses a flaw after disclosure. GTIG’s reported average of exploited zero-days rose from eight per month in 2025 to 11 per month in 2026, with 22 in August. Zero-days accounted for 62% of vulnerabilities exploited from January through August, according to the account. GTIG suggested that increased n-day exploitation may explain much of the broader growth in exploitation, so rising exploitation totals should not be interpreted as a rise in zero-days alone.
Rank #3
A rapid example: CVE-2026-1731
SecurityWeek reported that CVE-2026-1731, an unauthenticated OS command-injection flaw in BeyondTrust Privileged Remote Access and Remote Support, was autonomously discovered by the Hacktron AI research agent. GTIG reportedly observed one threat cluster exploit it within four days of public disclosure and five more within seven days. This is a specific reported case, not a measure of how quickly all vulnerabilities are exploited.
Are AI-discovered vulnerabilities more dangerous?
In GTIG’s reported comparison, 50% of vulnerabilities classified as likely AI-discovered enabled remote code execution, compared with 26% of vulnerabilities not so classified. GTIG also compared the groups using low-, medium-, and high-risk categories and observed exploitation status. The available coverage does not explain the classification method in enough detail to independently assess how vulnerabilities were assigned to either group.
The difference is notable, but it does not establish that AI alone caused the higher remote-code-execution share. The comparison is between groups GTIG classified, not a reported controlled causal study. GTIG suggested that AI models may help identify memory-corruption and logic flaws that traditional static analyzers miss; that explanation is a proposed interpretation, not proof of a mechanism.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problemsWhy might attackers move faster after disclosure?
Finding a new vulnerability and weaponizing a disclosed one are different activities. Once a patch, product-version change, disclosure announcement, or proof-of-concept is public, attackers may be able to analyze the difference and adapt existing techniques without discovering a new zero-day.
Best Value
SecurityWeek quotes GTIG as saying: “It is possible that threat actors are finding it more accessible or efficient to use LLMs and AI tools to automate analysis of differences between product versions, patches, vulnerability disclosure announcements, and Proof-of-Concept (POC) code to rapidly weaponize n-days, rather than to discover new zero-days.” The wording is explicitly a possibility offered by GTIG, not a demonstrated explanation for every observed attack.
AI-discovered flaws are not the same as flaws in AI systems
The report also tracked vulnerabilities in AI-related software, a separate category from vulnerabilities found with AI. SecurityWeek says GTIG tracked 2,076 AI-related CVEs from January 2025 through August 2026, including more than 1,500 in 2026; roughly half affected AI orchestration frameworks. Only a handful were confirmed exploited, and GTIG observed no zero-day exploitation of AI infrastructure during the period. These figures describe vulnerabilities in AI systems, not the share of all vulnerabilities discovered using AI.
What should security teams take from the findings?
The data supports treating public disclosure as a time-sensitive operational signal, without assuming every new CVE warrants the same response. The BeyondTrust example shows why teams should know which externally reachable products they run and be prepared to evaluate disclosures promptly.
Free tools Windows power users keep installed
One-click scans. No signup required.
- Keep an accurate inventory of deployed software, versions, and internet-facing assets.
- Monitor disclosures for products in that inventory, then check whether affected systems are exposed and whether exploitation has been reported.
- Prioritize patching using exposure, exploitability, business impact, and observed exploitation—not the raw number of CVEs alone.
These are practical implications of the reported trends and example, not operational instructions attributed to GTIG. GTIG expects vulnerability discovery and exploitation rates to continue increasing in the short to medium term, according to SecurityWeek; that is a forecast, not an observed result or a guarantee.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
