I'll be the first to admit I'm mediocre at exploit development, but I am in contact with enough extremely talented exploit developers to say this for certain:
Anyone claiming they were "easily" able to generate a working php2shell exploit with AI is either:
a) a world-class an extreme case of imposter syndrome.
b) currently looking at some very confident-sounding but wrong LLM output.
I'm rather certain it's the latter based on the number of AI-hallucinated exploits I've seen posted on GitHub. Threat actors are also blasting my honeypot network with "php2shell" exploits that not only don't work, but aren't even within 10,000 miles of anything to do with how the exploit actually works.
I've seen the same technique from the attached image in multiple alleged "php2shell exploits", including one on GitHub with over 100 stars. Problem is, it's not php2shell, nor is it even a vulnerability at all. It's a technique known as "session poisoning", which can be leveraged to exploit local file inclusion (LFI) vulnerabilities in PHP apps (not PHP itself).
The only thing it has in common with CVE-2026-109618/CVE-2026-109619 is they both use the "PHP_SESSION_UPLOAD_PROGRESS" feature. And mitigation advice for both issues reference setting "session.upload_progress.enabled" to "Off".
People are shoving loose details of the vulnerability into LLMs, the LLMs are keying in on the "PHP_SESSION_UPLOAD_PROGRESS" string, then generating session poisoning code that has nothing to do with the actual CVE. And because the people doing this have absolutely no idea what they're doing, they're confident enough to post their "proof of concept exploit" to GitHub and share it on social media.
So if you see people talking about "session poisoning + LFI" it's likely their LLM hallucinated an exploit based on a 2018 blog by Orange Tsai talking about a local file inclusion attack vector.
What's even worse is that "security researchers" are then using LLMs to triage these exploitation attempts, falling victim to the same hallucinations, then confirming there is ongoing mass-exploitation (there currently isn't).
The actual real exploit is very difficult to exploit, and so far I've not seen anyone successfully get it working on more than a handful of PHP version with highly-specific configurations.
LLMs are useful tools in the hands of skilled professionals, but for others, they're often just a Dunning-Kruger accelerator.
Anyone claiming they were "easily" able to generate a working php2shell exploit with AI is either:
a) a world-class an extreme case of imposter syndrome.
b) currently looking at some very confident-sounding but wrong LLM output.
I'm rather certain it's the latter based on the number of AI-hallucinated exploits I've seen posted on GitHub. Threat actors are also blasting my honeypot network with "php2shell" exploits that not only don't work, but aren't even within 10,000 miles of anything to do with how the exploit actually works.
I've seen the same technique from the attached image in multiple alleged "php2shell exploits", including one on GitHub with over 100 stars. Problem is, it's not php2shell, nor is it even a vulnerability at all. It's a technique known as "session poisoning", which can be leveraged to exploit local file inclusion (LFI) vulnerabilities in PHP apps (not PHP itself).
The only thing it has in common with CVE-2026-109618/CVE-2026-109619 is they both use the "PHP_SESSION_UPLOAD_PROGRESS" feature. And mitigation advice for both issues reference setting "session.upload_progress.enabled" to "Off".
People are shoving loose details of the vulnerability into LLMs, the LLMs are keying in on the "PHP_SESSION_UPLOAD_PROGRESS" string, then generating session poisoning code that has nothing to do with the actual CVE. And because the people doing this have absolutely no idea what they're doing, they're confident enough to post their "proof of concept exploit" to GitHub and share it on social media.
So if you see people talking about "session poisoning + LFI" it's likely their LLM hallucinated an exploit based on a 2018 blog by Orange Tsai talking about a local file inclusion attack vector.
What's even worse is that "security researchers" are then using LLMs to triage these exploitation attempts, falling victim to the same hallucinations, then confirming there is ongoing mass-exploitation (there currently isn't).
The actual real exploit is very difficult to exploit, and so far I've not seen anyone successfully get it working on more than a handful of PHP version with highly-specific configurations.
LLMs are useful tools in the hands of skilled professionals, but for others, they're often just a Dunning-Kruger accelerator.
Oct 23, 2026 · 2:53 AM UTC
135
572
3,575
340,633
I want to caveat this since based on some comments, people are interpreting it in a way other than was intended. You 10000% need to update your PHP version. php2shell is no joke.
I am NOT saying people don't have functional exploits...they do. I'm saying that most of the public ones are AI hallucinations, and there's also some complexities which lower the risk of something apocalyptic, like a self-propagating worm that blasts through the entire internet (À la. EnternalBlue or Log4J).
The chances are if you used some click-to-deploy container, installed your web stack from a tutorial, or copy and pasted a command from somewhere, you're probably running the same exact config as a gazillion other people. There is no guarantee someone won't publicly release an exploit that affects you setup, or multiple major setups. There is no way for you to know ahead of time.
Additionally, if you're running mod_php, that appears to be the easiest target of all. php-fpm and other configs are not immune, I've seen working exploits for all. But mod_php is the easiest, so I'd recommend not only updating PHP, but planning to migrate to something newer, like PHP-FPM.
I would bet good money this is not the last of the critical PHP bugs we're going to see, and the next ones could be easier to exploit. In which case you get less time to update.
Updating your PHP version is the absoloute bare minimum you NEED to do. Harden your servers, install security monitoring/detection tools, and please, please, migrate off mod_php if you can, it's obsolete af.
I am NOT saying people don't have functional exploits...they do. I'm saying that most of the public ones are AI hallucinations, and there's also some complexities which lower the risk of something apocalyptic, like a self-propagating worm that blasts through the entire internet (À la. EnternalBlue or Log4J).
The chances are if you used some click-to-deploy container, installed your web stack from a tutorial, or copy and pasted a command from somewhere, you're probably running the same exact config as a gazillion other people. There is no guarantee someone won't publicly release an exploit that affects you setup, or multiple major setups. There is no way for you to know ahead of time.
Additionally, if you're running mod_php, that appears to be the easiest target of all. php-fpm and other configs are not immune, I've seen working exploits for all. But mod_php is the easiest, so I'd recommend not only updating PHP, but planning to migrate to something newer, like PHP-FPM.
I would bet good money this is not the last of the critical PHP bugs we're going to see, and the next ones could be easier to exploit. In which case you get less time to update.
Updating your PHP version is the absoloute bare minimum you NEED to do. Harden your servers, install security monitoring/detection tools, and please, please, migrate off mod_php if you can, it's obsolete af.
3
15
210
12,826
@MalwareTechBlog yeah, hallucinated exploits all around. but once the exploit is public, spraying to other targets unlike before isn't hard.
x.com/S1r1u5_/status…
x.com/S1r1u5_/status…
i don't think there's a working public exploit. people seem to be just hallucinating with models, the bug is hard to exploit.
also, it's wildly irresponsible to publish anything about the bug before the embargo lifts. even hinting at the technical details is dangerous. please don't do it
2
0
52
18,797
@S1r1u5_ There was at least one working public exploit available that I've seen, but I've not observing any ITW exploitation at all. The current PoCs seem to be limited to specific PHP version with specific software installed and people are struggling to adapt them to be broadly usable
5
1
86
13,968
@MalwareTechBlog The existence of the exploit is already on the internet, the gun is assembled sir. Nobody using deepseek or glm is assembling the gun right now, they are pointing it at things. I like your tiktok content but this is you rn
6
0
15
12,331
@HtxYasi Sorry, but everyone with access to actual data disagrees with you. Not one vendor has confirmed successful exploitation.
3
1
164
8,340
@MalwareTechBlogand agree with this:
“LLMs are useful tools in the hands of skilled professionals, but for others, they're often just a Dunning-Kruger accelerator.”
when ppl say professionals are not required because ai can do this or that, my counter argument is to say “let me do it with ai and compare the results”
they will convince themselves why professionalism is still important ¯\_(ツ)_/¯
“LLMs are useful tools in the hands of skilled professionals, but for others, they're often just a Dunning-Kruger accelerator.”
when ppl say professionals are not required because ai can do this or that, my counter argument is to say “let me do it with ai and compare the results”
they will convince themselves why professionalism is still important ¯\_(ツ)_/¯
1
1
27
3,344
@MalwareTechBlog local llm blackhat hcker here!!! im doing my part!!! the script will be released to the public when i am done testing. 8-)
1
0
15
3,427
@MalwareTechBlogs only really feasible to exploit if you were using apache and mod_php because it involves causing a memory overflow on a session process and then reinjecting back malicious code onto that particular process.
With multiple workers running in an nginx + php-fpm configuration it becomes pot luck whether on not you get to pick up the same worker process that you overflowed. That’s why no exploits have been demonstrated against FPM for the second CVE.
With multiple workers running in an nginx + php-fpm configuration it becomes pot luck whether on not you get to pick up the same worker process that you overflowed. That’s why no exploits have been demonstrated against FPM for the second CVE.
2
1
14
3,784
@MalwareTechBlog It’s “difficult” to bootstrap from scratch. But you just need one basic exploit, and then it’s really just “now make it work for this Wordpress image”
4
0
78
7,259
@MalwareTechBlog Well yes but how we will sell AI doom with that attitude
1
1
27
3,664
@MalwareTechBlog the honeypot is getting spammed with fan fiction
0
0
20
3,887
@MalwareTechBlog is the whole php exploit just bs then? you mentioned mass exploitation is not happening. that has not been the case in the past with other "critical" vulnerabilities. do you think there will be a point where this exploit will have a proper poc made by a human and not an AI chud?
0
0
0
2
@MalwareTechBlog full RCE with ASLR bypass only work for php 8.4 (i confirm). Older versions of PHP require a different workaround to work (I am not familiar with them).
2
0
2
189
@MalwareTechBlog Have no idea what most of your acronyms stand for, but I recognize “Dunning-Kruger accelerator”. 😂
0
0
1
586
@MalwareTechBlog Luckily governments around the world employ teams of world class exploit developers. But at least John with Opus 5.5 can't touch this
0
0
0
452
@MalwareTechBlog @cyb3rops Can everyone please wake up and realize there’s a war going on? APTs are well funded and highly motivated.
They’re not hobbyists with a cushy private sector job doing it for funsies and clout on X.
Know your adversary & stop downplaying the severity of the situation.
They’re not hobbyists with a cushy private sector job doing it for funsies and clout on X.
Know your adversary & stop downplaying the severity of the situation.
0
0
0
18
@MalwareTechBlog seeing it in the wild too. the 100 star github poc is the 2018 orange tsai session poisoning trick with a php2shell readme on it. of the 90 odd repos claiming a poc, the ones that work need an lfi you brought yourself
0
0
1
849
@MalwareTechBlog I don’t understand a word any of you are saying. When yall run the world please treat me nice!
0
0
1
849
@MalwareTechBlog People lacking in core skills are using LLM w/o understanding what the LLM is outputting. They then glorify it as the ultimate superpower. "AI" companies lean in on those narratives to inflate their own "value". Shit always rises to the top, but only for as long as we allow it.
0
1
1
129
@MalwareTechBlog I can't make this exploit work. In fairness I'm the retard that runs the front of the house.
But also in fairness nobody in the back of the house has been able to show me this hotness working.
But also in fairness nobody in the back of the house has been able to show me this hotness working.
0
0
2
1,381
@MalwareTechBlog is that video of a guy “pwning” a bunch of sites also misconstrued? And if so, how. they make it look easy, looks terrifying
php2shell
Tomorrow is a helluva day.
0
0
0
355
@MalwareTechBlog FWIW the maintainers have been aware of the underlying primitive for 2 years and, due to technical challenges, left it unpatched because they believed that exploiting it was practically impossible. So... yeah.
0
0
0
354
@MalwareTechBlog The Problem is the tiny space to issue the payload, I had success writing a simple php cmd shell but it took hitting the host with 6 different requests writing the shell bit by bit
0
1
2
1,532
@MalwareTechBlog Watched a team lose a day to a PoC that was never real. Stars are not peer review
0
0
0
281
@MalwareTechBlog wasn't there a @tryhackme or @hackthebox_eu room for basic PHP shells that instructed users to make a script called php2shell.py2 to upload and exploit.
maybe some people got confused with the google results and are sending the tutorial script thinking its the new real deal :P
maybe some people got confused with the google results and are sending the tutorial script thinking its the new real deal :P
1
0
1
202
@MalwareTechBloghe potential possibilty of thousands of servers being hacked that worried people and the PHP guys wanted to make sure as many ppl as possible got it patched. Having read the docs etc think it was quite overblown as was a complicated/obscure exploit and would take a while for the script kiddies to build a replica of the alleged working exploit let alone get it working on real-world sites. Got to remember too, authorities & governments around the world now love to join in and overdramatise almost everything(it was the wind today here in the UK, no joke) just IN CASE something bad happens and then they can’t be blamed. The UK NCSC put out some huge notice about it telling ppl to put their sites offline even.
0
0
0
341
@MalwareTechBlog This bolsters my theory that simtelligent tools will only provide useful output for the intelligent, and garbage for the larpers
0
0
1
618
@MalwareTechBlog “but for others, they're often just a Dunning-Kruger accelerator” 👌🏽😂😂😂😂 hereinafter referred to as DKA
0
0
0
142
@MalwareTechBlog The more things change, the more the script kiddies stay the same. AI can’t change a scripty into a haxx0r.
0
0
0
280
@MalwareTechBlog Or what we've seen in the wild are fake exploits including a backdoor to infect the script kiddie or “AI hacker” device.
0
0
0
421
@MalwareTechBlog I'm not taking it lightly, but I've decided not to worry about it just yet with the sites I run, for the very reasons you mention. Nothing critical on these sites, so if they go they go, and I'll work on recoding them at my leisure over the next few weeks instead of rushing it.
0
0
1
3,044
@MalwareTechBlog I am also confident the whole thing is a) overblown because somebody (deliberately) connected it to AI, and b) I am pretty sure I heard about this exploit years ago already because my Ansible template contains a rule to disable X-header version to limit exposure, or I'm tripping.
0
0
0
343
@MalwareTechBlog exactly. it is a looong chain and requires a LOT of heap shaping. difficulty increases even more if you try it in a build with ASLR on
0
0
0
330
@MalwareTechBlog I'm just trying to put in gdb to see how it works but AI giving me the wrong answers that don't work and making doing all this stuff. Like WTF? oh right, all these AI CEO companies are lying piece of shit. Got it.
0
0
1
956
