Subscribe
Cybersecurity

Java4Shell probing climbs to 261 requests from 7 IPs, still no confirmed breach

Saturday, November 21, 2026 · 5 min read

GreyNoise counted 261 Java4Shell-shaped requests from 7 IP addresses by Nov. 21, up from 187 requests from 5 IPs as of Nov. 19 — but there is still no confirmed compromise, no public exploit that achieves code execution, and the flaw is not in CISA's KEV catalog.

Bearish$ORCL

📊 Scorecard: our Bearish call on $ORCL is scored from $101.19, the last close before it ran. Its first result comes after the next close.

Java4Shell probing climbs to 261 requests from 7 IPs, still no confirmed breach

The News

The knocking is getting louder, but nobody has broken in yet. Probing for Java4Shell, the Java flaw tracked as CVE-2026-114938, continues to climb: GreyNoise, a company that monitors internet scanning, counted 261 Java4Shell-shaped requests from 7 IP addresses by Nov. 21, up from 187 requests from 5 IPs as of Nov. 19. Probing means automated attempts to test whether a system is vulnerable — like someone trying door handles to see which are unlocked. GreyNoise reports zero successful intrusions on its sensors, so the attempts are being seen and counted, not succeeding.

The flaw is not in CISA's Known Exploited Vulnerabilities catalog as of the Nov. 20 catalog version, which means the US government has no confirmed case of it being used in a real attack. No public exploit achieves code execution — the ability to run an attacker's own code on a victim's system — either. The research firm Calif has published a trigger and a technical write-up, which show the flaw is real, but not a full exploit chain that turns it into a working attack. (Calif's researcher is Quang Luong.) That distinction matters: a trigger proves the bug exists, a working exploit proves it can be used, and so far only the first half exists in public.

The warnings are still escalating. Cybersecurity agencies in the US, Canada and South Korea jointly urged organisations to patch immediately — CISA's joint alert with Canada came on Nov. 17, and South Korea's science ministry and KISA urged emergency inspections on Nov. 18, as reported by Yonhap — a rare three-country warning that signals how seriously governments take a flaw in Java, one of the most widely used programming languages in the world. ThreatDefence's Nov. 21 update warns of fake proof-of-concept repositories circulating — files pretending to demonstrate the flaw that may themselves be malicious — and notes Red Hat products that bundle Java, such as Keycloak and JBoss EAP images, were still affected on Nov. 20. Oracle and OpenJDK fixed the flaw on Nov. 17 in builds 17.0.21.1, 21.0.13.1, 25.0.5.1 and 27.0.1.1. This post covers the Nov. 20–21 update; the Nov. 20 post covered the Nov. 19 probing counts and vendor impact lists.

Probing is not a breach — but 261 knocks on the door means the patch should already be in.

Why It Matters

The gap between probing and compromise is the whole story. Probing is the phase where attackers map out who is vulnerable; it is noisy, visible and, on its own, harmless. The danger is what comes next, when someone turns a published trigger into a working exploit and starts using it at scale. The jump from 187 to 261 requests in two days shows the scanning is accelerating, and the joint warning from three countries' agencies shows governments expect that escalation to continue.

The fake proof-of-concept warning is a second-order risk worth taking seriously. When a high-profile flaw is in the news, developers and security teams go looking for demonstration code to test whether they are affected. Attackers know this, and seed fake repositories that look like legitimate demonstrations but actually infect the people who run them. The safest way to test for Java4Shell is to use a trusted vendor's guidance, not a random repository found through a search.

The Red Hat detail shows why patching is not as simple as updating one program. Java is bundled inside many other products — identity systems like Keycloak, application servers like JBoss EAP — and each of those needs its own update. An organisation that patches its Java installation but forgets the products that ship their own copy is still exposed, which is why the vendor lists matter as much as the headline fix.

Analyst's Take

Our call is Bearish. A flaw in one of the world's most widely used programming languages, with probing climbing, three governments urging immediate patching and fake exploits already circulating, is bad news for Oracle's Java ecosystem and for every company running Java in production. The saving grace is that there is still no confirmed breach — but that grace period is exactly what the patching window is for.

What to watch next is whether a working public exploit appears and whether the flaw lands in CISA's KEV catalog. Either event would change the story from a watch-and-patch situation into an active-incident one, and would likely trigger a fresh wave of emergency patching. Until then, the numbers to track are GreyNoise's probing counts, which have risen on every update so far.

For organisations running Java, the message is simple and unchanged: patch now, check the products that bundle Java, and do not download proof-of-concept code from untrusted sources. Reader poll: Have you patched Java yet — yes or no? Vote: Patched, Patching soon, Not affected.

Are you bullish or bearish on this story?

By The Numbers

261Java4Shell-shaped requests counted by GreyNoise by Nov. 21
7IP addresses the requests came from
0Successful intrusions on GreyNoise's sensors
187Java4Shell-shaped requests counted as of Nov. 19
3Countries whose cybersecurity agencies jointly urged patching — the US, Canada and South Korea
Nov. 17Date Oracle and OpenJDK fixed the flaw

FAQ

What is Java4Shell?

Java4Shell is the nickname for CVE-2026-114938, a security flaw in Java, one of the most widely used programming languages in the world. The name echoes Log4Shell, a famous Java flaw from 2021. Like Log4Shell, it could allow remote code execution — running an attacker's code on a victim's system — if a working exploit is developed.

What is a proof of concept?

A proof of concept, often shortened to PoC, is demonstration code that shows a security flaw is real by triggering it in a controlled way. Security researchers publish them so defenders can test whether they are affected. Attackers also publish fake ones, which is why the source of the code matters as much as the code itself.

What is CISA's KEV catalog?

It is a list kept by the US Cybersecurity and Infrastructure Security Agency of security flaws known to be actively exploited in the real world. A flaw's absence from the list does not mean it is safe — it means no confirmed exploitation has been recorded yet. Java4Shell is not on the list as of the Nov. 20 catalog version.

What does remote code execution mean?

Remote code execution means an attacker can run their own code on a victim's computer or server from across the internet, without needing a login. It is one of the most serious kinds of security flaw, because once an attacker can run code, they can usually steal data, install malware or take over the system entirely.

👁

Source: ThreatDefence

Get the daily briefing. AI news with a markets lens — tickers, bullish/bearish reads, zero hype. Subscribe free on Substack

More on $ORCL

All $ORCL coverage →

More from AI Agent Update