Corroboration Monitor · No. 1
Half of Stories Had a Second Organization in Three Days; One in Five Had None
The first Corroboration Monitor follows each activity from its first public report to the day an independent organization reported it too. It counts stories, Asterism's clusters of reports, so a report nobody else wrote about is outside the curve, and in this window that was most of them.

Half of the stories first reported between 19 Sep and 31 Oct 2026 had a second, independent organization reporting them within 2 to 3 days. The count covers 92 stories from 134 organizations. By 13 Nov, 20% of the stories (95% CI 13% to 29%) still had only one.
Key Findings
- The median wait for a second organization was 2 days (95% CI 1 to 5). Reasonable changes to the sample keep it within 2 to 3 days.
- On the day of the first report, 26% of the stories already had a second organization, and 54% had one within three days.
- By the as-of date, at least 14 days after their first report, 20% (95% CI 13% to 29%) had no second organization.
- Stories that a vendor reported first were corroborated more often (87%, n = 61) than those that began on a community list (63%, n = 19), though the intervals overlap.
- Stories are the minority: 82% of the window's 2,908 reports joined no story at all.
Days to a Second Organization
- second organization
- third organization
- shaded band, the interval
- tick, still waiting at the as-of date
- median
as of 13 Nov 2026Measures reporting, not activity.Download CSV
Data table survival.csv
| group | days | s | lo | hi | censored |
|---|---|---|---|---|---|
| second organization | 0 | 0.739 | 0.637 | 0.817 | 0 |
| second organization | 1 | 0.554 | 0.447 | 0.649 | 0 |
| second organization | 2 | 0.500 | 0.394 | 0.597 | 0 |
| second organization | 3 | 0.457 | 0.353 | 0.554 | 0 |
| second organization | 4 | 0.402 | 0.302 | 0.500 | 0 |
| second organization | 5 | 0.359 | 0.262 | 0.456 | 0 |
| second organization | 6 | 0.283 | 0.195 | 0.377 | 0 |
| second organization | 7 | 0.272 | 0.186 | 0.365 | 0 |
| second organization | 8 | 0.239 | 0.158 | 0.330 | 0 |
| second organization | 9 | 0.207 | 0.131 | 0.294 | 0 |
| second organization | 10 | 0.196 | 0.122 | 0.282 | 0 |
| second organization | 14 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 15 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 16 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 22 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 28 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 29 | 0.196 | 0.122 | 0.282 | 2 |
| second organization | 31 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 33 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 38 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 42 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 44 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 45 | 0.196 | 0.122 | 0.282 | 3 |
| second organization | 47 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 49 | 0.196 | 0.122 | 0.282 | 1 |
| second organization | 53 | 0.196 | 0.122 | 0.282 | 1 |
| third organization | 0 | 0.880 | 0.795 | 0.932 | 0 |
| third organization | 1 | 0.750 | 0.648 | 0.826 | 0 |
| third organization | 2 | 0.739 | 0.637 | 0.817 | 0 |
| third organization | 3 | 0.728 | 0.625 | 0.807 | 0 |
| third organization | 4 | 0.717 | 0.613 | 0.798 | 0 |
| third organization | 5 | 0.707 | 0.602 | 0.788 | 0 |
| third organization | 6 | 0.696 | 0.591 | 0.779 | 0 |
| third organization | 8 | 0.685 | 0.579 | 0.769 | 0 |
| third organization | 14 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 15 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 16 | 0.685 | 0.579 | 0.769 | 3 |
| third organization | 17 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 19 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 20 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 21 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 22 | 0.685 | 0.579 | 0.769 | 4 |
| third organization | 23 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 24 | 0.685 | 0.579 | 0.769 | 5 |
| third organization | 25 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 26 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 28 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 29 | 0.685 | 0.579 | 0.769 | 5 |
| third organization | 30 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 31 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 32 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 33 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 36 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 37 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 38 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 39 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 41 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 42 | 0.685 | 0.579 | 0.769 | 3 |
| third organization | 43 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 44 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 45 | 0.685 | 0.579 | 0.769 | 3 |
| third organization | 47 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 49 | 0.685 | 0.579 | 0.769 | 3 |
| third organization | 50 | 0.685 | 0.579 | 0.769 | 1 |
| third organization | 52 | 0.685 | 0.579 | 0.769 | 2 |
| third organization | 53 | 0.685 | 0.579 | 0.769 | 4 |
The teal curve falls fast: on day zero, the day of the first report, 26% of the stories already had a second organization, and by day seven only 27% were still waiting. After two weeks it flattens at 20%: a story that hadn't been picked up by then mostly wasn't picked up at all.
A third organization is rarer: only 29 of the 92 stories reached one, and that curve never falls to one half, so there's no median to give.
On 13 Nov, 18 stories were still waiting. The curve counts each of them up to that day and no further, which is why we use a survival curve: dropping them would make the waits look shorter than they were.
Corroboration by Who Reported First
- whisker, the interval
as of 13 Nov 2026Measures reporting, not activity.Download CSV
Data table shares.csv
| category | share | lo | hi | n |
|---|---|---|---|---|
| vendor | 0.869 | 0.762 | 0.932 | 61 |
| community | 0.632 | 0.410 | 0.809 | 19 |
| government or CERT | 0.750 | 0.468 | 0.911 | 12 |
Asterism records a category for each source, and these are the categories of the first publisher. Vendor-first stories were corroborated 87% of the time (95% CI 76% to 93%), government and CERT-first stories 75% (47% to 91%), and community-first stories 63% (41% to 81%). Two of the three groups are small samples, and all three intervals overlap, so read this as a lean, not a ranking.
The community group is mostly one place: the oss-security mailing list, where open-source maintainers and researchers post vulnerability disclosures, was first in 17 of its 19 stories. Not every open-source flaw gets a vendor write-up, which may explain the lower share.
Each Story's Wait
- median
as of 13 Nov 2026Measures reporting, not activity.Download CSV
Data table lags.csv
| category | days | story |
|---|---|---|
| community | 0 | |
| community | 0 | |
| community | 0 | |
| community | 1 | |
| community | 1 | |
| community | 1 | |
| community | 1 | |
| community | 2 | |
| community | 6 | |
| community | 6 | |
| community | 9 | |
| community | 10 | |
| government or CERT | 0 | |
| government or CERT | 0 | |
| government or CERT | 0 | |
| government or CERT | 0 | |
| government or CERT | 1 | |
| government or CERT | 1 | |
| government or CERT | 1 | |
| government or CERT | 1 | |
| government or CERT | 4 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 0 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 1 | |
| vendor | 2 | |
| vendor | 2 | |
| vendor | 2 | |
| vendor | 2 | |
| vendor | 3 | |
| vendor | 3 | |
| vendor | 3 | |
| vendor | 3 | |
| vendor | 4 | |
| vendor | 4 | |
| vendor | 4 | |
| vendor | 4 | |
| vendor | 5 | |
| vendor | 5 | |
| vendor | 5 | |
| vendor | 5 | |
| vendor | 6 | |
| vendor | 6 | |
| vendor | 6 | |
| vendor | 6 | |
| vendor | 6 | |
| vendor | 7 | |
| vendor | 8 | |
| vendor | 8 | |
| vendor | 8 | |
| vendor | 9 | |
| vendor | 9 |
Most dots sit between zero and three days, and 7 stories waited more than a week for their second organization, the longest 10 days.
Most Reports Stand Alone
- in a story
- standing alone
as of 13 Nov 2026Measures reporting, not activity.Download CSV
Data table weekly.csv
| week | kind | reports |
|---|---|---|
| 2026-09-21 | in a story | 55 |
| 2026-09-21 | standing alone | 322 |
| 2026-09-28 | in a story | 59 |
| 2026-09-28 | standing alone | 387 |
| 2026-10-05 | in a story | 48 |
| 2026-10-05 | standing alone | 370 |
| 2026-10-12 | in a story | 84 |
| 2026-10-12 | standing alone | 427 |
| 2026-10-19 | in a story | 183 |
| 2026-10-19 | standing alone | 483 |
| 2026-10-26 | in a story | 94 |
| 2026-10-26 | standing alone | 386 |
A story needs two reports about the same activity. Of the window's 2,908 dated intelligence reports, 524 joined a story and 82% (95% CI 81% to 83%) did not. Some of those are one-off research nobody needed to repeat; some are activities a second organization covered under different enough details that the clustering missed the pair. They are outside the curves above, so the curves describe activities that drew at least two reports, not everything published.
The jump in the week of 19 Oct is PHP2Shell, the pair of PHP flaws disclosed on 21 Oct. Within a fortnight, 89 organizations had reported it, in one story.
Limits of This Report
- This measures reporting, not activity. A fast second report means a second organization published, not that it saw the activity itself. Corroboration is not confirmation: a second organization can repeat the first one's mistake.
- Independent means a different organization, not independent evidence. In our validation sample, 3 of the 18 second reports added nothing of their own: an article cross-posted on two blogs, a joint advisory published by both of its authors, and a vendor's summary of a government alert. Two of those pairs may be one publisher under two names; counting each as one gives a median of 3 days, and 23% of stories with no second organization.
- "First report" is the first in Asterism's panel, from publishers that allow AI use. Someone outside the panel may have reported it earlier.
- Dates are whole days, so a same-day second report is a tie, not a race anyone won.
- Recent stories had less time. A first report from late October that gets its second report in mid-November isn't a story yet, so it is missing rather than censored. Limiting the sample to first reports up to 15 Oct, which had at least 29 days, gives a median of 3 days (n = 56).
- The clustering makes mistakes. We corrected one (see the method) and left one: PHP2Shell is split across two stories. Folding the second into the first gives a median of 3 days.
- The panel grew from 183 sources reporting in the window's first two weeks to 196 in its last two, so later stories had a few more potential witnesses.
- No source in the window carries Asterism's news category, so this edition can't separate press coverage from original research.
- We don't rank named publishers by speed, and the data files don't either: they list who reported what, and when.
Method
The definitions of a report, a story, an organization, a first report and corroboration are those of How Null Sec Counts. The population is every story whose first report is dated 19 Sep to 31 Oct 2026 (UTC), as Asterism's stories stood on 13 Nov. The window starts on 19 Sep, the first whole day after Asterism's coverage date. For each story we took each source's earliest report in it, mapped the source to its organization, and measured the days from the first report to the second and third organizations' first reports.
Asterism's MCP server names the organization only for a story's first source, so we mapped the other sources by name, in organizations.csv. Our count of organizations matched Asterism's for every story but two. In one, we count two JPCERT/CC feeds as one publisher where Asterism counts two; Asterism's count gives the same median of 2 days. In the other, PHP2Shell, we left out Arete's "September Threat Trends" page.1 It is dated 15 Oct, a week before the disclosure, but it is a podcast page that later gained articles about events of 21 Oct, so its date isn't when it reported PHP2Shell. With Asterism's dates as they stand, the median is 3 days. The next report in that story, Blackpoint Cyber's, opened on 20 Oct with the PHP team's advance notice, and that date stands.2
The curves are Kaplan-Meier estimates with Greenwood intervals on a log-log scale, and the median's interval is where the band crosses one half. Shares have Wilson intervals. We checked the median three ways decided before looking at the data: first reports up to 15 Oct only (3 days, n = 56); without the single most frequent first publisher (2 days, n = 75); and without news organizations as witnesses, which didn't apply because there were none. Three more came out of reading the data: Asterism's dates as they stand (3 days), the PHP2Shell stories folded into one (3), and the suspected same-publisher pairs counted once (3). The results run from 2 to 3 days, so that range is the headline.
To check the clustering, we drew 20 corroborated stories at random and compared the first report with the second organization's. Of those, 18 described the same activity. Of the other two, one was the PHP2Shell story, whose misdated first report we corrected; in the other, both reports concern one CVE, but the first covers the fix and the second a later backport for older devices. The sample and our reading of each pair are in validation.csv, and every query is in queries.aql.
About the Data
- As of
- Window
- 19 Sep to 31 Oct 2026
- Coverage since
- Panel
- 183 sources reported in the first two weeks of the window, 196 in the last two; the data counted here come from 134 organizations
- Marking
- TLP:CLEAR
- Left out
- Undated reports, marketing and product news, round-ups (never clustered), publishers who opted out of AI use, and anything above TLP:CLEAR. Single reports that joined no story are outside the survival analysis.
This measures reporting, not activity. It counts what was published.
Null Sec sells Asterism, the platform these data come from.
Data and Queries
The files behind every figure and number in this report, as published. Each SHA-256 lets you check you have the same bytes.
- survival.csvThe data behind Fig. 1.2.7 KB · SHA-256
021fba30df95a797de20e6db8dc4b69893f02040377425c98ad14dbe3afe8891 - shares.csvThe data behind Fig. 2.122 bytes · SHA-256
bd6ef4dc59079b88e320f1584cdd530000448f4cdce2ec01cde8d908a22d9b48 - lags.csvThe data behind Fig. 3.905 bytes · SHA-256
f27c104d4a9b6ceb2b6a62f9d0dfe953449cded72b4945abab3f46f6e2710f12 - weekly.csvThe data behind Fig. 4.349 bytes · SHA-256
1195fa832e5577702b20926b48400049f32b09ac7924160bd804573183198dc9 - organizations.csvSupporting data.11.9 KB · SHA-256
4f8b8bfb6b7ae8ee2b52c23d8bce4b05742abe2c0e79e2a1b01fa52ac980d191 - stories.csvSupporting data.6.0 KB · SHA-256
5f141bbdc17e9d5b6a48f81562a8cc761d2cd3d3a0101f7dea24c28955e408b8 - story-sources.csvSupporting data.45.0 KB · SHA-256
c14c68633049665ff12137ec59ac2880f7ad4f5daed798cfa11c02a1ef9a276b - validation.csvSupporting data.8.7 KB · SHA-256
c64e1423439dd050a7337d95351a926f8edccb849f63324ae9ed2c32f2258c2c - queries.aqlThe exact queries, with their windows and the dates they ran.19.5 KB · SHA-256
3e0ecf9c9a8756a0e421622891df55070fd130dbaec806f5f9dd1105c4159f87 - sources.csvEvery report cited or counted, with its publisher, date, URL and TLP.453 bytes · SHA-256
80e7538f923c0c566d6b5ce3a2c047bfd88def3704d0169770020ee2bd4948c6
References
- Arete, "Cyber Campfire: September Threat Trends & Insights", dated 15 Oct 2026 in Asterism. areteir.com/resources/cyber-campfire-september-threat-trends-insights ↑
- Blackpoint Cyber, "PHP2Shell: Critical PHP RCE CVE-2026-109618 & CVE-2026-109619", opened 20 Oct 2026. blackpointcyber.com/blog/php2shell-cve-2026-109618-cve-2026-109619/ ↑
Spot an error? Write to us and we'll fix it here, with a dated note.