Skip to content
Private Beta ·invite-only access. Reach out to get in.

Detector · headline threats · last 21 days

What is actually burning right now

Only the vulnerabilities the security press is reporting on widely: named by at least 4 of the sources we read within 14 days, with articles about that vulnerability from at least 2 different news outlets. Shown for 21 days after the coverage last met that bar. A vulnerability scored below 8.0 that is not known to be exploited is left out. Everything else is left out, whatever its score.

Exploited

CVE-2026-123890

9.8CVSS

VMware vCenter

Reported by 10 security news and advisory sources since 2026-11-19.

AI summary of the linked sources, written 2026-11-22

An attacker can obtain an authenticated administrator session on VMware vCenter Server.

Impact:
An attacker can obtain an authenticated administrator session on VMware vCenter Server.
Ease of exploit:
The flaw is exploitable remotely without authentication; Google GTIG reported that Armory Castle has been exploiting the vCenter zero-day since at least early September.
Root cause:
An unauthenticated Secure Remote Password (SRP-6a) session key disclosure over Distributed Computing Environment/Remote Procedure Calls (DCE/RPC) that results in authentication bypass.
Fix:
Install Broadcom fixes for vCenter in versions 9.1.1.0200, 9.0.2.0200, or 8.0 U3m.

headline since 2026-11-20 · published 2026-11-19

  • Network & cloud
  • KEV
  • POC

Radar shows all 6

  • Inner zone: headline and known to be exploited (CISA KEV, or CISA reports active exploitation)
  • Outer zone: headline, not known to be exploited
  • Nearer the centre: became headline more recently (approximate)
  • Slices, by the kind of system our analysis gave each CVE: Applications 2 · Web 2 · Systems 0 · Network & cloud 1 · Hardware & devices 0 · Other & unclassified 1
  • Dot size and colour: CVSS score (below 7.0, 7.0–8.9, 9.0+, hollow: no score)
6Headline threats
4Known to be exploited
78%Of CVSS 8.0+ CVEs do not meet our exploitability rule

News scan as of 2026-11-22 12:48 UTC. Exploit evidence as of 2026-11-22 13:50 UTC, refreshed about every 5 minutes.

Headline threats

The 6 on the radar

6 shown

Known to be exploited · all 4 shown

Exploited10 sources

CVE-2026-123890

9.8CVSS

VMware vCenter

Reported by 10 security news and advisory sources since 2026-11-19.

AI summary of the linked sources, written 2026-11-22

An attacker can obtain an authenticated administrator session on VMware vCenter Server.

Impact
An attacker can obtain an authenticated administrator session on VMware vCenter Server.
Ease of exploit
The flaw is exploitable remotely without authentication; Google GTIG reported that Armory Castle has been exploiting the vCenter zero-day since at least early September.
Root cause
An unauthenticated Secure Remote Password (SRP-6a) session key disclosure over Distributed Computing Environment/Remote Procedure Calls (DCE/RPC) that results in authentication bypass.
Fix
Install Broadcom fixes for vCenter in versions 9.1.1.0200, 9.0.2.0200, or 8.0 U3m.

Sources. Summary written from: helpnetsecurity.com, thehackernews.com, securityweek.com. Other coverage: cybersecuritynews.com, theregister.com, bleepingcomputer.com.

  • Network & cloud
  • KEV
  • POC
  • CVSS 9.8

headline since 2026-11-20 · published 2026-11-19

Exploited7 sources

CVE-2026-109618

9.2CVSS

PHP

Reported by 7 security news outlets since 2026-10-23.

AI summary of the linked sources, written 2026-11-22

PHP 8.2 before 8.2.35, 8.3 before 8.3.36, 8.4 before 8.4.27 and 8.5 before 8.5.12 have an out-of-memory handling bug that can allow remote code execution.

Impact
Allows remote code execution against the affected PHP process.
Ease of exploit
Unauthenticated RCE over HTTP has been demonstrated against Apache mod_php.
Root cause
PHP's out-of-memory abort handling (longjmp) can interrupt initialization leaving incomplete request state reachable, and request shutdown may access that invalid state.
Fix
Upgrade to PHP 8.2.35, 8.3.36, 8.4.27, or 8.5.12, or a later release in the corresponding branch.

Sources. Summary written from: openwall.com, php2shell.delftintellab.com, openwall.com. Other coverage: thehackernews.com, cybersecuritynews.com, infosecurity-magazine.com, securityweek.com, thehackernews.com, bleepingcomputer.com.

  • Applications
  • KEV
  • EXPLOIT
  • POC
  • EPSS P99
  • CVSS 9.2

headline since 2026-10-28 · published 2026-10-21

Exploited6 sources

CVE-2026-109619

9.2CVSS

PHP

Reported by 6 security news outlets since 2026-10-23.

AI summary of the linked sources, written 2026-11-22

PHP contains a stale global state use-after-free that can allow remote code execution.

Impact
A use-after-free in PHP can allow remote code execution.
Ease of exploit
Unauthenticated RCE has been demonstrated against Apache mod_php and PHP-FPM, and exploitability depends on the affected state, configuration, and application behavior.
Root cause
PHP maintains global state that temporarily points into request memory which can be accessed after the memory is freed, causing a use-after-free.
Fix
Upgrade to PHP 8.2.35, 8.3.36, 8.4.27, or 8.5.12; updating is the only reliable mitigation.

Sources. Summary written from: openwall.com, arcticwolf.com, openwall.com. Other coverage: thehackernews.com, cybersecuritynews.com, infosecurity-magazine.com, thehackernews.com, bleepingcomputer.com, cybersecuritynews.com.

  • Web
  • KEV
  • EXPLOIT
  • POC
  • EPSS P99
  • CVSS 9.2

headline since 2026-10-28 · published 2026-10-21

Exploited4 sources

CVE-2026-115731

9.4CVSS

Perforce Puppet Enterprise

Reported by 4 security news and advisory sources since 2026-11-18.

AI summary of the linked sources, written 2026-11-22

Perforce Puppet Enterprise deserialization of untrusted data in the report-processing code allows a client with any CA-signed agent certificate to execute arbitrary code on the Puppet Server as the service user.

Impact
An attacker can execute malicious payloads on the server with elevated privileges and could expose the server's certificate-signing key to generate trusted certificates and alter software or configurations.
Root cause
Deserialization of untrusted data in Puppet's report-processing code.
Fix
Puppet Enterprise and Puppet Enterprise Advanced versions 2026.0.1, 2025.12.1, and 2023.8.13; Puppet Core versions 9.1.3 and 8.22.2.

Sources. Summary written from: news4hackers.com, thehackernews.com, helpnetsecurity.com. Other coverage: wid.cert-bund.de.

  • Other & unclassified
  • KEV
  • POC
  • CVSS 9.4

headline since 2026-11-22 · published 2026-11-03

Not known to be exploited · all 2 shown

Headline13 sources

CVE-2026-114938

8.1CVSS

Oracle Java SE

Reported by 13 security news and advisory sources since 2026-11-18.

AI summary of the linked sources, written 2026-11-21

OpenJDK's Base64 decoder has a memory-corruption flaw that allows an unauthenticated attacker to achieve remote code execution.

Impact
An attacker can overwrite heap memory, read it, and reach remote code execution.
Ease of exploit
Requires supplying specially crafted Base64 input and can allow unauthenticated remote code execution.
Root cause
A bug in OpenJDK's optimized machine-level Base64 decoding code for x86-64 allows malformed input to overwrite neighboring memory.
Fix
Start OpenJDK with either -XX:-UseBASE64Intrinsics or -XX:UseAVX=2; apply this flag only on x86-64 hosts.

Sources. Summary written from: openwall.com, calif.io, cisa.gov. Other coverage: theregister.com, news4hackers.com, news4hackers.com, infosecurity-magazine.com, thehackernews.com, therecord.media.

  • Applications
  • POC
  • EPSS P98
  • CVSS 8.1

headline since 2026-11-18 · published 2026-11-17

Headline9 sources

CVE-2026-123007

9.2CVSS

F5 NGINX Plus

Reported by 9 security news and advisory sources since 2026-11-18.

AI summary of the linked sources, written 2026-11-22

The most impactful of the five findings, CVE-2026-123007, allows an unauthenticated attacker to execute commands on an affected nginx server with HTTP/2 enabled.

Impact
The resulting memory corruption can disclose data, interrupt service, and execute code with the nginx worker process's permissions.
Ease of exploit
Leveraging CVE-2026-123007 for RCE requires a significant amount of network traffic to be transmitted, approximately 128 GB of HTTP/2 data, and available network bandwidth and connection timeouts place practical limitations on Internet exploitation.
Root cause
HTTP/2 flow-control memory corruption caused by an integer underflow.
Fix
NGINX Plus 37.1.0 - 37.2.0.1 37.3.0.1; NGINX Plus 37.0.0 - 37.0.7.1 37.0.8.1; NGINX Plus R34 - R36 None; NGINX Open Source 1.31.0 - 1.31.8 1.31.9; NGINX Open Source 1.9.5 - 1.30.6 1.30.7

Sources. Summary written from: rapid7.com, xint.io, cyberscoop.com. Other coverage: theregister.com, thehackernews.com, helpnetsecurity.com, infosecurity-magazine.com, bsi.bund.de, cert.europa.eu.

  • Web
  • POC
  • EPSS P98
  • CVSS 9.2

headline since 2026-11-20 · published 2026-11-18

010078%594 of 766

Noise meter · CVSS 8.0+ · last 14 days

Bullshit meter

This is a different measure from the radar above. Of the 766 CVEs scored 8.0 or higher that our pipeline analysed in the last 14 days, 594 (78%) do not meet our exploitability rule: none of them is known to be exploited or has a working public exploit, and each one either needs something an outside attacker does not have (a login, a user action, local access, special conditions) or is scored below 9.0 with no public proof-of-concept and no high EPSS rank.

Scope: CVEs our pipeline analysed, not every CVE published. Older CVEs we imported in bulk are left out. The score is the published CVSS; the split is our own rule, read from the CVSS vector and public exploitation evidence. It is not an industry measure, and it says nothing about any vendor.

Meets our exploitability rule
172
22%

Does not meet it, by first blocker

Local or adjacent access only
50
7%
Needs a login
247
32%
Needs a user action
184
24%
High complexity or special conditions
57
7%
Scored below 9.0, no exploitation evidence
56
7%

Each CVE is counted once, by the first that applies in this order: access, login, user action, complexity. Counted from the CVSS vector. Rows add up to 594.

172 meet the rule. Meeting it does not make a CVE a headline threat: the radar shows only what the press reports widely.

How the detector decides

What counts as a headline threat

Within 14 days, at least 4 different sources name the CVE in an article that names at most 5 CVEs, and at least 2 different news outlets publish an article about that CVE itself, not a roundup. We read about 35 news, CERT and vendor feeds three times a week. This follows the news: most of these were already being exploited when the coverage arrived. It is not an early warning.

Where the summary comes from

Written by a language model from the pages linked under each card as “Summary written from”, and only from them, on the date shown. It is not updated when the pages change. A point the pages do not support is left out. It can still be wrong: read the sources.

What this page does not cover

Anything the press does not write about widely, however severe: a CVE can be exploited and never appear here. Coverage in languages and outlets we do not read. “Known to be exploited” means listed in CISA's Known Exploited Vulnerabilities catalog or reported by CISA as actively exploited; a public exploit alone is shown as a mark, not as exploited. The slice is the kind of system our own analysis gave the CVE. The sweep is decoration.