GTIG
Google's threat intelligence team, responsible for collecting and analyzing global cybersecurity threat data and publishing security research reports.
Timeline (last 90 days)
5,045 vulnerabilities were disclosed in January 2026, and by August it had become 10,740, doubling.
Actually exploited vulnerabilities account for only 0.23% of all disclosed vulnerabilities, roughly only 1 out of 431.
GTIG points out that automated CNA numbering policies in the open source ecosystem push up the baseline numbers, and the number of IDs reflects the aggressiveness of the process rather than an expansion of the real attack surface.
High-risk disclosures soared from 131 in January 2026 to 350 in August, a 167% increase, but still accounted for only 3% of all disclosures.
In 2025 an average of 10.5 vulnerabilities were exploited per month, rising to 18 per month from January to August 2026.
The main source of the growth in vulnerability exploitation from January to August 2026 was concentrated in the rapid weaponization of n-days, with zero-day exploitation rising only slightly from 8 to 11 per month, while high-risk exploitation doubled from 28 in 2025 to 75 in the first eight months of 2026.
Zero-days still account for 62% of exploited vulnerabilities, and since May 2026 the growth in exploitation (indexed +127%) has moved in sync with the growth in disclosures (indexed +128%).
GTIG speculates that AI tools have made n-day weaponization cheaper and more efficient, with threat actors using LLMs and AI tools to automatically analyze product version differences, patches, vulnerability disclosure advisories, and PoC code to rapidly weaponize n-days, which is easier than discovering new zero-days.
Among the vulnerabilities GTIG could identify as possibly AI-discovered, low-risk findings dropped to 39% (69% for traditional findings), medium-risk rose to 58% (28% for traditional), and exactly 50% of AI-discovered vulnerabilities led to remote code execution, versus 26% across the entire ecosystem.
Between January 2025 and August 2026, a cumulative 2,076 AI-related CVEs were disclosed, of which more than 1,500 came in 2026 alone, concentrated in orchestration frameworks (782, half of all AI-related flaws, a 347% surge) and inference/serving infrastructure (212).
2 more timeline events
All Facts (14)
GTIG是谷歌在2024年整合旗下多个安全团队后形成的统一威胁情报力量，融合了Mandiant、VirusTotal和TAG的能力
90%UnverifiedGTIG拥有数百名安全分析师和研究人员，覆盖全球超过100个国家的威胁监测网络
75%Unverified5,045 vulnerabilities were disclosed in January 2026, and by August it had become 10,740, doubling.
50%UnverifiedActually exploited vulnerabilities account for only 0.23% of all disclosed vulnerabilities, roughly only 1 out of 431.
50%UnverifiedGTIG points out that automated CNA numbering policies in the open source ecosystem push up the baseline numbers, and the number of IDs reflects the aggressiveness of the process rather than an expansion of the real attack surface.
50%UnverifiedHigh-risk disclosures soared from 131 in January 2026 to 350 in August, a 167% increase, but still accounted for only 3% of all disclosures.
50%UnverifiedIn 2025 an average of 10.5 vulnerabilities were exploited per month, rising to 18 per month from January to August 2026.
50%UnverifiedThe main source of the growth in vulnerability exploitation from January to August 2026 was concentrated in the rapid weaponization of n-days, with zero-day exploitation rising only slightly from 8 to 11 per month, while high-risk exploitation doubled from 28 in 2025 to 75 in the first eight months of 2026.
50%UnverifiedZero-days still account for 62% of exploited vulnerabilities, and since May 2026 the growth in exploitation (indexed +127%) has moved in sync with the growth in disclosures (indexed +128%).
50%UnverifiedGTIG speculates that AI tools have made n-day weaponization cheaper and more efficient, with threat actors using LLMs and AI tools to automatically analyze product version differences, patches, vulnerability disclosure advisories, and PoC code to rapidly weaponize n-days, which is easier than discovering new zero-days.
50%UnverifiedAmong the vulnerabilities GTIG could identify as possibly AI-discovered, low-risk findings dropped to 39% (69% for traditional findings), medium-risk rose to 58% (28% for traditional), and exactly 50% of AI-discovered vulnerabilities led to remote code execution, versus 26% across the entire ecosystem.
50%UnverifiedBetween January 2025 and August 2026, a cumulative 2,076 AI-related CVEs were disclosed, of which more than 1,500 came in 2026 alone, concentrated in orchestration frameworks (782, half of all AI-related flaws, a 347% surge) and inference/serving infrastructure (212).
50%UnverifiedVulnerabilities in edge and security devices account for 14% of exploited vulnerabilities, more than 65% of exploited edge flaws reach a high/critical threat risk rating, and adversaries actively target unauthenticated public management interfaces.
50%UnverifiedGTIG recommends shifting from unprioritized mass patching to threat-intelligence-driven triage and handling, combining targeted edge defense with automated, agentic remediation, and conducting internal pre-release review of AI code.
50%