We seeded 2m canaries into popular AI services and tracked where they were stolen and abused.
This identified:
- Three compromised providers
- Resellers stealing prompts via built-in logging
- Providers ignoring ZDR & privacy flags
https://t.co/ySVOzX8SUd
Pre-auth RCE in nginx is the bug every hacker dreams of finding. It almost never happens.
Today weโre disclosing nginx2root: five nginx vulnerabilities, and an exploit that takes an unauthenticated attacker all the way to root.
One of themhits the default config.
If you run nginx, patch now. Then read: https://t.co/rDQQAimLZP
CVE-2026-123007, CVE-2026-123634, CVE-2026-123633, CVE-2026-121810, CVE-2026-113425
Looks like Java4Shell will be a more limited affair, sourcing details from the @datadoghq writeup by @Frichette_n and Sebastian Obregoso:
1. Requires x86-64 CPU with AVX-512 VBMI. From what I can gather, this is most newer generation cloud instances that arenโt ARM. You can run `grep -om1 -E 'avx512[_]?vbmi' /proc/cpuinfo` on the host to see if youโre potentially impacted.
2. Itโs far easier to try and abuse this to cause a DoS than a full RCE. RCE must be targeted and is not generic.
3. Patches are out and two mitigations exist: starting the JVM with option `-XX:-UseBASE64Intrinsics` or a more breaking change `-XX:UseAVX=2` which reduces performance and will break ARM64 JVMs if itโs in a shared config.
4. The full exposure is really JDK 17.0.6+ on AVX-512 VBMI x86 hosts, with code that Base64-decodes untrusted input through the JDKโs basic or URL decoders.
Iโd treat this as a high-priority DoS with a targeted RCE possibility, not a Log4Shell.
๐จWe urge orgs to immediately patch the recently disclosed Oracle Java vuln CVE-2026-114938. This critical memory-corruption vulnerability, known as โJava4Shell,โ affects the Base64 decoder of the x86-64 HotSpot JVM, part of the Java Development Kit. ๐https://t.co/SwrhdPFlZF
GreyNoise is tracking #java4shell (CVE-2026-114938) with a live dashboard here: https://t.co/cZ7tOYOvk2
At this time, we are not aware of successful exploitation in the wild. We believe this will be challenging to mass exploit at least initially due to the need for the exploit to be tailored for each targeted system build.
@duniel_pls I can dig further in a little bit and see what I can come up with.
(also, this reminds me I never forwarded the ClickFix intel I said I would!) ๐คฆโโ๏ธ
Still have a nagging feeling that there is more to the story... I have been going through other sources of intel and figured BGP would be interesting to look at. There are a few odd announcement spikes preceding the FAA ground stop, but nothing definitive to link them together๐ค
@Andrew___Morris Yeah I have too many IR engagements in flight right now to have the time to dig in like I used to... Just the spidey sense is saying something is off..
watchTowr Intel is aware of public reports of an unauthenticated RCE in Java, CVE-2026-114938, affecting OpenJDK 17.0.6 and later. Details are still unconfirmed. Review your Java environments now and monitor for vendor announcements.
Hereโs everything you need to know about Java2Shell (CVE-2026-114938). You probably shouldnโt ignore this one.
Apparently, thereโs a pre-auth RCE affecting that tiny application nobody uses, Java.
Early reports suggest OpenJDK 17.0.6+ is affected, with a 9.2 CVSS 4.0 score.
Before testing all AI slop scripts from on GitHub, know that there is NO public PoC out yetโฆ
Weโre still waiting for the full technical details, including the actual exploit conditions. In the meantime, probably a good idea to figure out where Java is running in your environment.
https://t.co/L7RoadAHfR
You wouldn't believe the number of networks I've seen saved by canary credentials, this stuff works and it works extremely well.
The first thing most threat actors do when they successfully break into a network is look for useful credentials to pivot off of or escalate access with. Make sure they find some.
fake API keys, usernames and passwords, session tokens, NTLM hashes, password.txt. Put them everywhere. Servers, employee endpoints, password managers, the web browser's password/session databases, cloud storage buckets, network shares.
The credentials should look legitimate, high value, and be unique to each individual system or user. That way you can set up granular detection for what happens when an attacker attempts to log in with a canary credential.
You can not only instantly lock out the system it was stolen from, but also any real credentials belonging to that system/users, which may have been exfiltrated in the process.
@kmcnam1 As the world's most unlucky traveler who has slept overnight on many airport floors due to missed flights (also a former medsec network engineer who gets deployment timing), I feel am qualified to speak on the subject. 12am is absolutely a better time to deploy core changes...
Day 2 of the Microsoft outage. I have spent the past 3 years reporting on Microsoft and have written extensively about the companyโs security failures. As we plow into the AI age, Iโm still reporting. If you have relevant information to share, reach me on Signal at 929-317-0748.
This OpenSSH auth bypass vulnerability is crazy!
Under the right conditions, attackers can bypass SSH authentication for any account that meets the requirements.
Prerequisites:
โ musl libc (e.g. Alpine Linux, Openwrt)
โ OpenSSH without PAM
โ SSH password authentication enabled
โ Target account has * in /etc/shadow โ Another account has a valid password hash
As an example, right now, official Alpine Linux cloud images on AWS and Azure have a default account that can run commands as root without a password. Those cloud images are vulnerable to this attack.
https://t.co/GGsB1E5nae