Security
The 2026 DBIR in One Sentence: Attackers Stopped Needing Your Password
By VirtuesTech Team ·
For most of the DBIR's nineteen years, stolen credentials were the front door. This year they aren't. In the 2026 report, exploitation of vulnerabilities is the most common initial access vector for breaches, at 31%, while credential abuse fell to 13%.
The dataset behind that is large: more than 31,000 incidents, over 22,000 of them confirmed breaches, across 145 countries. So it is worth asking what it changes in a testing plan. We think three things.
1. The patch queue is now the attack surface
The report's most uncomfortable number is about remediation. Only 26% of critical vulnerabilities, meaning those in CISA's Known Exploited Vulnerabilities catalog, were fully remediated in 2025, down from 38% the year before. Median time to full resolution went from 32 days to 43, and the median organization had 50% more critical vulnerabilities to patch.
Read together, that is a queue growing faster than teams can clear it, in a year when exploitation became the leading way in.
What helps is knowing which items in the queue an attacker can actually reach. On our own exposure platform, one capture shows 38,221 raw scanner hits collapsing to 450 deduplicated issues before anything is graded. Most of a raw queue is noise, and the work that matters is finding the part that isn't.
In a testing plan: rank findings by demonstrated exploitability, and test the fix. A patch that was applied but never verified is still an open item, as far as an attacker is concerned.
2. Your vendors are half the story
Breaches with third-party involvement rose 60% on the previous year's dataset and now make up 48% of the total. The report also looked at how third parties handle their own cloud exposure: only 23% fully remediated missing or weak multifactor authentication on their cloud accounts.
In a testing plan: test the paths a vendor's compromise would open into your environment, and ask your vendors for their own recent test evidence. If a vendor can't show a recent, retested report, treat that as a finding in its own right.
3. People are still in most breaches, and the phone is the new inbox
The human element was present in 62% of breaches, up slightly from 60%. In phishing simulations, the median click rate on mobile-centric lures, such as voice and text messages, ran 40% higher than on email. Pretexting reached 6% of breaches.
In a testing plan: a vulnerability list will never show you this. An objective-driven red team will, because it goes where a real intruder would, including through people, and then shows which of those steps your monitoring saw.
What we'd change this quarter
- Retest every critical fix. The gap between "patched" and "verified" is where the 26% figure lives. We include one retest cycle in every engagement for exactly this reason.
- Rank by exploitability. A CVSS score tells you how bad a flaw could be. Whether it is reachable in your environment decides whether it is urgent.
- Bring vendors into scope. Map which third parties can reach what, and test those paths.
- Test detection, not just prevention. If ransomware is in 48% of breaches, the question is how quickly you would notice.
The report's own summary of the year is "Refinement, not revolution." We agree. None of this needs a new program, but it does need a tighter loop between finding, fixing, and proving the fix.
Figures are quoted from the Verizon 2026 Data Breach Investigations Report Executive Summary, published May 2026.