Most disclosed software vulnerabilities never appear in real-world attacks, but the ones that do tend to give attackers control of a system. Microsoft’s analysis points defenders towards prioritising vulnerabilities that enable remote code execution, authentication bypass or privilege escalation, where delays can leave critical systems exposed for months.
Watch Desk analysis
What happened
In a security analysis published on 2 November, Microsoft says roughly 383,000 vulnerabilities had been disclosed from 1999 to September 2026, but only about 2,000, or 0.5%, were linked to real-world exploitation. It says takeover vulnerabilities account for 78% of exploited vulnerabilities in its analysis.
The mix differs from the broader CVE population: Microsoft reports that server-side remote code execution makes up 37% of exploited vulnerabilities, compared with 20% of all published CVEs. Its 2026 patching data also show a wide gap by system type: more than a month for Windows servers, over ten weeks for Linux servers, and more than a year for network appliances at the median.
Why it matters
When teams cannot patch everything at once, the findings offer a practical way to prioritise: consider what a flaw lets an attacker do, how reachable the affected system is, and what that system could expose if compromised. Microsoft argues that impact and exposure together help identify vulnerabilities with the greatest potential blast radius.
The timing is uncomfortable. Microsoft says some high-impact flaws are being weaponised within days of disclosure, and cites industry research suggesting AI advances are a significant factor in faster exploit development. That is the company’s account of the trend, not a measure here of how much faster AI makes exploitation.
Our read
The useful message is not that every vulnerability deserves the same emergency response. It is that patch queues should reflect the route into a network and the damage a successful attack could do, not just a score attached to a CVE. Microsoft sells security tools alongside this analysis, so treat its product recommendations separately from the data and prioritisation argument.
What to watch
- Whether other research supports the reported patching delays and exploitation patterns.
- How defenders weigh exposure and asset importance alongside vulnerability severity.
- Whether evidence clarifies AI’s contribution to faster exploit development.
Discussion spark: Should organisations prioritise patching by a flaw’s likely blast radius and exposure, even when that means leaving higher-scoring vulnerabilities waiting?
Sources and evidence
- What exploited vulnerabilities have in common: A data-driven analysis of real-world attacks – Microsoft (2 November 2026, 20:20 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.