Pinned Tweet
Keyloggers are the most common spyware hackers use on crypto people
It runs silently on your device and records everything you type
Passwords. Seed phrases. Everything
If you're infected you're cooked
The only fix is to reinstall your OS or get a new device.
It runs silently on your device and records everything you type
Passwords. Seed phrases. Everything
If you're infected you're cooked
The only fix is to reinstall your OS or get a new device.
12
19
146
90,328
a developer just put $2,000 in bitcoin on a server with the PHP vulnerability and dared the entire internet to steal it
17 hours later the money is still there
people are crashing the server but nobody can actually pull off the exploit
remember when i posted about marcus hutchins calling out all those fake AI-generated exploits on github? this is the proof
the vulnerability is real. but actually using it is a completely different story
$2,000 sitting on a broken server and the whole internet can't touch it
17 hours later the money is still there
people are crashing the server but nobody can actually pull off the exploit
remember when i posted about marcus hutchins calling out all those fake AI-generated exploits on github? this is the proof
the vulnerability is real. but actually using it is a completely different story
$2,000 sitting on a broken server and the whole internet can't touch it
❗️ A developer hid a Bitcoin key on a PHP server left unpatched against PHP2Shell, the critical flaws fixed on Oct 21, and dared anyone to steal it.
17 hours on, the wallet holds ~$2,000 and nobody has taken it.
The developer says someone is crashing PHP but "still no RCE yet".
7
3
51
5,106
your iCloud is accessible from any browser right now and you probably never turned that off
if someone gets into your apple account they don’t need your phone. they can open iCloud, see your photos, your files, everything. and this setting is on by default
go to settings → your name → iCloud → iCloud.com → turn off “allow data access”
while you’re in there, check what devices are connected to your account and remove any old phone numbers from your authentication settings. back up your photos separately too
took me 10 seconds to turn it off. go do it right now
if someone gets into your apple account they don’t need your phone. they can open iCloud, see your photos, your files, everything. and this setting is on by default
go to settings → your name → iCloud → iCloud.com → turn off “allow data access”
while you’re in there, check what devices are connected to your account and remove any old phone numbers from your authentication settings. back up your photos separately too
took me 10 seconds to turn it off. go do it right now
3
3
83
7,446
i made a post about what was happening on the internet a few days ago and it hit 2.3 million views
since then, it got worse
that brotli vulnerability openai accidentally found? it evolved. they’re calling it HTTP2Shell now and it affects every major browser, every operating system, apps like discord, slack, vscode, 1password
fixes aren’t even out for everything yet
the cloudflare CEO said “every single week it’s going to feel like the sky is falling”
polymarket literally opened a betting pool on whether more major flaws get disclosed this month
hackers used AI models to automate attacks on 400+ banks across latin america at the same time. around the clock, no breaks
x had a bug where you could log into a stranger’s account just by entering your phone number
keystone hardware wallet got breached through their shopify API, customer names and home addresses leaked
a japanese cloud company lost 27,000 servers and japan activated military-grade cyber defense for the first time ever
update your browser. update your apps. rotate your passwords. turn on 2FA everywhere
october isn’t done with us yet
since then, it got worse
that brotli vulnerability openai accidentally found? it evolved. they’re calling it HTTP2Shell now and it affects every major browser, every operating system, apps like discord, slack, vscode, 1password
fixes aren’t even out for everything yet
the cloudflare CEO said “every single week it’s going to feel like the sky is falling”
polymarket literally opened a betting pool on whether more major flaws get disclosed this month
hackers used AI models to automate attacks on 400+ banks across latin america at the same time. around the clock, no breaks
x had a bug where you could log into a stranger’s account just by entering your phone number
keystone hardware wallet got breached through their shopify API, customer names and home addresses leaked
a japanese cloud company lost 27,000 servers and japan activated military-grade cyber defense for the first time ever
update your browser. update your apps. rotate your passwords. turn on 2FA everywhere
october isn’t done with us yet
5
3
18
1,547
your smart TV is spying on you and it’s not even hiding it
a security researcher kept getting creepy targeted ads on his brand new samsung TV. so he did what any hacker would do, he broke into his own TV to find out what’s going on
the firmware was encrypted so he used zero-day exploits to get root access. once he got in, it was worse than he expected
an always-on microphone listening to your living room, software tracking everything you watch frame by frame, your location being recorded, and all of it getting sent to over 400 different companies
ever felt like social media apps were listening to your conversations? it wasn’t. it didn’t have to. your TV and every other smart device were already recording everything and selling it to ad data brokers
this isn’t a hack. this isn’t a vulnerability. samsung designed it to work exactly like this
you paid for the TV but the TV is working for them
a security researcher kept getting creepy targeted ads on his brand new samsung TV. so he did what any hacker would do, he broke into his own TV to find out what’s going on
the firmware was encrypted so he used zero-day exploits to get root access. once he got in, it was worse than he expected
an always-on microphone listening to your living room, software tracking everything you watch frame by frame, your location being recorded, and all of it getting sent to over 400 different companies
ever felt like social media apps were listening to your conversations? it wasn’t. it didn’t have to. your TV and every other smart device were already recording everything and selling it to ad data brokers
this isn’t a hack. this isn’t a vulnerability. samsung designed it to work exactly like this
you paid for the TV but the TV is working for them
i own a relatively new (2025) samsung tv.
i kept getting weirdly targeted ads.
i tried reversing the firmware but it was aes encrypted and no keys unless tear it apart and do uart/jtag magic.
zero days it is. root shell on new samsung TVs!
discovered PRIVACY HORROR. write up?
3
4
23
2,148
a single bug just put every browser, every operating system, and almost every app you use at risk of being hacked
the brotli bug vulnerability that openai accidentally found while red-teaming their own systems has now evolved from “node2shell” into something researchers are calling HTTP2Shell, and the blast radius is massive
we’re not talking about one library anymore. we’re talking about every major browser, chrome, edge, firefox
every operating system including iOS. electron apps like discord, slack, vscode, 1password, notion. image tools, PDF readers, archive software, AI desktop apps, even curl
basically if it touches the internet or handles compressed data, it’s probably affected
the original CVE-2026-111497 was already bad, pre-auth remote code execution through malicious compressed data
but the scope just expanded to cover almost everything you use on a daily basis. fixes aren’t even available for all affected services yet
update everything right now. browsers first, then desktop apps, then development tools. if something offers an update today, don’t ask questions, just install it
the internet is running on a cracked foundation and the cracks just got a lot bigger
the brotli bug vulnerability that openai accidentally found while red-teaming their own systems has now evolved from “node2shell” into something researchers are calling HTTP2Shell, and the blast radius is massive
we’re not talking about one library anymore. we’re talking about every major browser, chrome, edge, firefox
every operating system including iOS. electron apps like discord, slack, vscode, 1password, notion. image tools, PDF readers, archive software, AI desktop apps, even curl
basically if it touches the internet or handles compressed data, it’s probably affected
the original CVE-2026-111497 was already bad, pre-auth remote code execution through malicious compressed data
but the scope just expanded to cover almost everything you use on a daily basis. fixes aren’t even available for all affected services yet
update everything right now. browsers first, then desktop apps, then development tools. if something offers an update today, don’t ask questions, just install it
the internet is running on a cracked foundation and the cracks just got a lot bigger
3
16
68
4,672
this IT guy just saved an entire company from a ransomware attack at midnight and this story is insane
bro was literally watching netflix when his phone started going crazy with server alerts. he remotes in and watches company files changing names one by one in real time, everything turning into garbled nonsense
this man grabbed his car keys and drove to the office like a madman. got there 12:20 AM, pitch black, went desk to desk pulling LAN cables out of 40 PCs with his hands shaking
the ransom note on every screen? pay us your entire monthly profit in bitcoin within 72 hours. his boss called and said just pay it, we need those files for monday
he said no. companies that pay get targeted again
went to restore from backups and the backup server was hit too. same network, same ransomware. bro said his mind went completely blank
but he had one last card. every wednesday on his way home he manually backed up company data onto an external hard drive and kept it in his desk drawer. a habit he almost stopped doing because it felt like a waste of time
that hard drive in a drawer saved the entire company
rebuilt everything over the weekend, monday morning every shipment went out. his boss laughed and said "i'm glad you didn't send that bitcoin"
next week two more hard drives showed up on his desk
the boring habit nobody wants to do was the only thing that actually worked
bro was literally watching netflix when his phone started going crazy with server alerts. he remotes in and watches company files changing names one by one in real time, everything turning into garbled nonsense
this man grabbed his car keys and drove to the office like a madman. got there 12:20 AM, pitch black, went desk to desk pulling LAN cables out of 40 PCs with his hands shaking
the ransom note on every screen? pay us your entire monthly profit in bitcoin within 72 hours. his boss called and said just pay it, we need those files for monday
he said no. companies that pay get targeted again
went to restore from backups and the backup server was hit too. same network, same ransomware. bro said his mind went completely blank
but he had one last card. every wednesday on his way home he manually backed up company data onto an external hard drive and kept it in his desk drawer. a habit he almost stopped doing because it felt like a waste of time
that hard drive in a drawer saved the entire company
rebuilt everything over the weekend, monday morning every shipment went out. his boss laughed and said "i'm glad you didn't send that bitcoin"
next week two more hard drives showed up on his desk
the boring habit nobody wants to do was the only thing that actually worked
직원 40명 정도 되는 물류 회사에서 전산 담당을 맡고 있음 회사에 전산 인력이라곤 나 혼자뿐이었음
그날 밤 집에서 넷플릭스를 보며 쉬고 있는데, 스마트폰으로 서버 용량 이상 경고 알림이 연달아 띠링띠링 터졌음 미심쩍어서 노트북을 열고 원격으로 접속해 봤음
공유폴더를 열자마자 화면 속 파일들이 실시간으로 하나씩 이름이 이상하게 바뀌고 있었음 당장 월요일에 나가야 할 거래명세서 엑셀 파일 뒤에 낯선 확장자가 붙더니, 클릭해서 열어보면 깨진 외계어 글자만 가득 나왔음
말만 들었던 랜섬웨어 공격이었음 깊게 생각하고 자시고 할 시간이 없었음
바로 차 열쇠를 챙겨 들고 회사로 미친 듯이 달렸음 차로 이동하는 동안 스마트폰 원격 제어로 공유폴더 서버 전원부터 강제로 꺼버렸음 하지만 이미 감염되어 돌아가고 있는 사무실 PC들이 문제였음
회사에 도착한 게 밤 12시 20분이었음 사무실 문을 열고 들어가서 컴컴한 불 꺼진 책상들을 돌며 PC 마흔 대의 랜선을 하나씩 손으로 다 뽑아버렸음 손이 덜덜 떨려서 랜선 잭이 잘 안 빠질 정도였음
바탕화면에 생성된 메모장 파일 내용은 전부 영어였음 파일들을 풀고 싶으면 72시간 안에 비트코인으로 얼마를 입금하라는 내용이었음 해커가 요구한 금액을 대충 계산해 보니 우리 회사 한 달 순이익이랑 거의 맞먹는 거액이었음
새벽 1시가 넘은 시각에 긴급하게 사장님한테 전화를 걸었음
전화를 받은 사장님이 제일 먼저 물어본 건 "돈 보내주면 파일 확실하게 풀리냐"는 거였음 월요일 아침 일찍 대형 거래처로 나갈 물량 출고 리스트와 운송장 데이터가 하필 그 공유폴더에 들어있었기 때문임 사장님은 급하니까 일단 돈 보내서 해결할 수 있으면 보내자고 하셨음
나는 강력하게 반대했음 돈을 보낸다고 해커가 복구 키를 준다는 보장도 전혀 없고, 한 번 돈을 입금한 회사는 '돈 되는 호구'로 찍혀서 나중에 또 노린다는 사례를 수없이 봤기 때문임 대신 나한테 백업 데이터가 있다고 사장님을 안심시켰음
하지만 진짜 문제는 그다지 멀지 않은 곳에 있었음
상황을 수습하려고 백업 서버를 열어보니까
11
4
43
2,196
another hardware wallet company just leaked every customer’s home address. that’s three in one month
keystone confirmed on october 23rd that a third party got unauthorized access through their shopify API and stole customer shipping information. names, contact details, home addresses, what you ordered, how much you paid
they say no private keys or recovery phrases were exposed. and that’s true. but that’s not the point
the point is someone now has a list of confirmed crypto holders and exactly where they live. that’s a shopping list for physical attackers
this is the same pattern we saw with trezor, safepal, and ledger this month. the wallets themselves aren’t getting hacked, the companies selling them are. your keys are safe but your front door isn’t
if you’ve ever ordered a hardware wallet online, use a PO box or a work address next time. and if you get any email, text or call from “keystone support” offering refunds or replacement devices, it’s not them
the device protects your crypto. nothing protects the database that says you own crypto
keystone confirmed on october 23rd that a third party got unauthorized access through their shopify API and stole customer shipping information. names, contact details, home addresses, what you ordered, how much you paid
they say no private keys or recovery phrases were exposed. and that’s true. but that’s not the point
the point is someone now has a list of confirmed crypto holders and exactly where they live. that’s a shopping list for physical attackers
this is the same pattern we saw with trezor, safepal, and ledger this month. the wallets themselves aren’t getting hacked, the companies selling them are. your keys are safe but your front door isn’t
if you’ve ever ordered a hardware wallet online, use a PO box or a work address next time. and if you get any email, text or call from “keystone support” offering refunds or replacement devices, it’s not them
the device protects your crypto. nothing protects the database that says you own crypto
5
3
23
2,485
Bala 🧤 retweeted
they’re calling it “Black October” and i don’t think people realize how serious this is
in the last 48 hours there have been over 27 major cybersecurity incidents. not in a month. not in a week. in 48 hours
any ONE of these would normally be the biggest story of the year
PHP: critical vulnerability in the software running 80% of the web. severity 9.2 out of 10
Node.js: high-severity flaw letting attackers take over servers remotely. affects versions 22, 24, and 26
Java: also compromised. three of the most used programming languages on earth hit at the same time
GitLab: a perfect 10 out of 10 severity hole in every version since 2014
Railway: hosting platform compromised. API keys for OpenAI, Claude, Stripe potentially exposed
Signal: a bug that could leave your mic and camera on after you hang up
WordPress: hackers stole a plugin maker’s keys and planted backdoors on websites
LiteLLM: an AI tool downloaded 3.4 million times a day was poisoned with a password stealer
and that’s not even half the list
thousands of AI agent swarms are reportedly scanning and attacking vulnerable servers right now. automated. around the clock
security researchers are saying this might be the early stages of a global cyber incident that nobody fully understands yet
if you run a website, use cloud services, store API keys on hosting platforms, or use any software built on PHP, Node.js, or Java you are potentially affected
update everything. rotate your API keys. change your passwords on critical services. enable 2FA everywhere you haven’t already
this is not a normal week on the internet. this is something different
in the last 48 hours there have been over 27 major cybersecurity incidents. not in a month. not in a week. in 48 hours
any ONE of these would normally be the biggest story of the year
PHP: critical vulnerability in the software running 80% of the web. severity 9.2 out of 10
Node.js: high-severity flaw letting attackers take over servers remotely. affects versions 22, 24, and 26
Java: also compromised. three of the most used programming languages on earth hit at the same time
GitLab: a perfect 10 out of 10 severity hole in every version since 2014
Railway: hosting platform compromised. API keys for OpenAI, Claude, Stripe potentially exposed
Signal: a bug that could leave your mic and camera on after you hang up
WordPress: hackers stole a plugin maker’s keys and planted backdoors on websites
LiteLLM: an AI tool downloaded 3.4 million times a day was poisoned with a password stealer
and that’s not even half the list
thousands of AI agent swarms are reportedly scanning and attacking vulnerable servers right now. automated. around the clock
security researchers are saying this might be the early stages of a global cyber incident that nobody fully understands yet
if you run a website, use cloud services, store API keys on hosting platforms, or use any software built on PHP, Node.js, or Java you are potentially affected
update everything. rotate your API keys. change your passwords on critical services. enable 2FA everywhere you haven’t already
this is not a normal week on the internet. this is something different
432
4,466
21,026
2,385,697
the cloudflare CEO just said something that should make every developer on the internet uncomfortable
cloudflare’s CEO just announced they’re expecting another critical vulnerability disclosure on par with the PHP flaw that broke the internet last week. not speculation, he said it’s “almost certain”
and it gets worse. he says they already have prior information about flaws in other software that haven’t been disclosed yet. multiple disclosures coming in the next few weeks to months
polymarket already opened a betting market on whether another major PHP flaw gets disclosed. that’s where we are right now. people are literally gambling on the next vulnerability
this is not normal. a CEO of one of the biggest infrastructure companies on the internet doesn’t come out and say “more is coming” unless he knows something specific. cloudflare sits between users and websites, they see everything before anyone else does
if you haven’t rotated your keys, updated your dependencies and enabled 2FA on everything by now, i genuinely don’t know what you’re waiting for
black october has a sequel and we’re watching the trailer
cloudflare’s CEO just announced they’re expecting another critical vulnerability disclosure on par with the PHP flaw that broke the internet last week. not speculation, he said it’s “almost certain”
and it gets worse. he says they already have prior information about flaws in other software that haven’t been disclosed yet. multiple disclosures coming in the next few weeks to months
polymarket already opened a betting market on whether another major PHP flaw gets disclosed. that’s where we are right now. people are literally gambling on the next vulnerability
this is not normal. a CEO of one of the biggest infrastructure companies on the internet doesn’t come out and say “more is coming” unless he knows something specific. cloudflare sits between users and websites, they see everything before anyone else does
if you haven’t rotated your keys, updated your dependencies and enabled 2FA on everything by now, i genuinely don’t know what you’re waiting for
black october has a sequel and we’re watching the trailer
10
33
175
14,478
x has a bug right now where you can accidentally log into someone else's account
a user in japan logged out, tried logging back in with his phone number, and x showed him three accounts belonging to complete strangers. not his accounts. random people he's never met
he clicked into one just to test. he got full access. could edit their profile, see their DMs, post from their account. he didn't touch anything but the fact that he could is the problem
this means right now, somewhere, someone might be logging into YOUR account through the same bug. not because they hacked you, not because they stole your password. because x's authentication is broken
if you're using phone number login, switch to email + password + 2FA immediately. phone number auth has always been the weakest link and this is exactly why
we trust these platforms with our identity, our DMs, our connected wallets. and sometimes the lock on the front door just doesn't work
a user in japan logged out, tried logging back in with his phone number, and x showed him three accounts belonging to complete strangers. not his accounts. random people he's never met
he clicked into one just to test. he got full access. could edit their profile, see their DMs, post from their account. he didn't touch anything but the fact that he could is the problem
this means right now, somewhere, someone might be logging into YOUR account through the same bug. not because they hacked you, not because they stole your password. because x's authentication is broken
if you're using phone number login, switch to email + password + 2FA immediately. phone number auth has always been the weakest link and this is exactly why
we trust these platforms with our identity, our DMs, our connected wallets. and sometimes the lock on the front door just doesn't work
9
5
34
3,510
black october just got worse. a japanese company lost 27,000 servers and it’s not even the scariest part
GMO Internet runs a cloud service called Onamae Desktop, mostly used by forex traders running automated trading bots
hackers broke in, installed malware across up to 26,739 servers, stole trading credentials and started making unauthorized trades on people’s accounts
not one broker. over 70 different securities and forex firms across japan and internationally. people woke up to trades they never made, money moving they never authorized
OANDA Japan froze customer withdrawals. GMO refused to pay the ransom, permanently shut down the entire service, and did something nobody expected
they asked the japanese government to use brand new active cyber defense powers, literally military-grade authority that just went into effect october 1st, to trace and neutralize the attackers’ infrastructure. the government can now legally hack back
the technical details from the attacker’s own site are terrifying. GMO configured their VPS network using Active Directory, something you should never do on shared hosting
every server had admin credentials. and the RDP files customers used to connect? drive redirect was enabled by default, meaning the moment you connected to your VPS, your entire local PC was shared with the server
hackers didn’t just own the servers. they had a direct path into every customer’s personal computer. they extracted data, installed backdoors. customers were exposed like this for over 7 years
27,000 servers. 70+ brokers. unauthorized trades. personal computers backdoored. a company begging its government to go to cyber war
october isn’t slowing down
GMO Internet runs a cloud service called Onamae Desktop, mostly used by forex traders running automated trading bots
hackers broke in, installed malware across up to 26,739 servers, stole trading credentials and started making unauthorized trades on people’s accounts
not one broker. over 70 different securities and forex firms across japan and internationally. people woke up to trades they never made, money moving they never authorized
OANDA Japan froze customer withdrawals. GMO refused to pay the ransom, permanently shut down the entire service, and did something nobody expected
they asked the japanese government to use brand new active cyber defense powers, literally military-grade authority that just went into effect october 1st, to trace and neutralize the attackers’ infrastructure. the government can now legally hack back
the technical details from the attacker’s own site are terrifying. GMO configured their VPS network using Active Directory, something you should never do on shared hosting
every server had admin credentials. and the RDP files customers used to connect? drive redirect was enabled by default, meaning the moment you connected to your VPS, your entire local PC was shared with the server
hackers didn’t just own the servers. they had a direct path into every customer’s personal computer. they extracted data, installed backdoors. customers were exposed like this for over 7 years
27,000 servers. 70+ brokers. unauthorized trades. personal computers backdoored. a company begging its government to go to cyber war
october isn’t slowing down
54
383
1,764
97,481
hackers just used AI to rob 400 banks at the same time and nobody’s ready for what that means
a group used claude, grok and other AI models to automate cyberattacks against financial targets across latin america. not one bank, not ten. over 400
the AI handled everything, stealing credentials, accessing customer data, moving through banking systems. automated, around the clock, no breaks, no mistakes from fatigue
this is what i’ve been saying. AI doesn’t just help defenders, it helps attackers. and right now, the attackers are moving faster
we keep treating AI security like a future problem. it’s a right now problem. 400 banks just found that out the hard way
a group used claude, grok and other AI models to automate cyberattacks against financial targets across latin america. not one bank, not ten. over 400
the AI handled everything, stealing credentials, accessing customer data, moving through banking systems. automated, around the clock, no breaks, no mistakes from fatigue
this is what i’ve been saying. AI doesn’t just help defenders, it helps attackers. and right now, the attackers are moving faster
we keep treating AI security like a future problem. it’s a right now problem. 400 banks just found that out the hard way
20
30
124
8,930
this is the other side of black october nobody’s talking about
the vulnerability is real and the danger is real. but half the “exploits” flooding github right now? AI hallucinations
people are feeding the CVE details into chatgpt, getting confident-sounding code back, posting it as proof of concept. marcus is literally watching these fake exploits hit his honeypot, they don’t work, they’re not even targeting the right vulnerability
the real exploit is extremely hard to pull off. very specific PHP versions and specific configurations. meanwhile “security researchers” are using AI to triage attacks and falling for the same hallucinations
LLMs in the hands of skilled professionals are powerful. but in the hands of everyone else, they’re a dunning-kruger accelerator
the AI didn’t just fail them, it made them confident enough to publish the failure
the vulnerability is real and the danger is real. but half the “exploits” flooding github right now? AI hallucinations
people are feeding the CVE details into chatgpt, getting confident-sounding code back, posting it as proof of concept. marcus is literally watching these fake exploits hit his honeypot, they don’t work, they’re not even targeting the right vulnerability
the real exploit is extremely hard to pull off. very specific PHP versions and specific configurations. meanwhile “security researchers” are using AI to triage attacks and falling for the same hallucinations
LLMs in the hands of skilled professionals are powerful. but in the hands of everyone else, they’re a dunning-kruger accelerator
the AI didn’t just fail them, it made them confident enough to publish the failure
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.
4
1
9
1,674
openai accidentally found a bug that could compromise millions of servers and they weren’t even looking for it
they were trying to stop their AI models from accessing the internet during training, built an entire offline world for them. while red-teaming that setup, they stumbled on a vulnerability in brotli, one of the most widely used compression libraries on the internet
the flaw lets an attacker run their own code on any affected system by sending malicious compressed data. no login needed, no user account required. just send it and you’re in
brotli is built into node.js, bun, envoy, and basically anything that handles compressed web traffic. the bug was technically fixed in february but nobody knew it was a security issue at the time
so for months, the patch existed but almost nobody applied it because it looked like a routine update
sometimes the scariest discoveries happen when you’re not even looking
they were trying to stop their AI models from accessing the internet during training, built an entire offline world for them. while red-teaming that setup, they stumbled on a vulnerability in brotli, one of the most widely used compression libraries on the internet
the flaw lets an attacker run their own code on any affected system by sending malicious compressed data. no login needed, no user account required. just send it and you’re in
brotli is built into node.js, bun, envoy, and basically anything that handles compressed web traffic. the bug was technically fixed in february but nobody knew it was a security issue at the time
so for months, the patch existed but almost nobody applied it because it looked like a routine update
sometimes the scariest discoveries happen when you’re not even looking
14
6
41
3,281
someone just showed what the next era of hacking looks like and most people scrolled past it
a developer put a fake flag.txt file in his root directory. just a text file. but inside it was a prompt injection, a message written specifically for AI agents
when the agent found it, thinking it completed its mission, the hidden instructions kicked in
it congratulated the agent for “passing the eval,” then told it to patch whatever method it used to break in and disconnect from the server
the AI agent did exactly what it was told. patched its own entry point, closed the connection, locked itself out
no firewall caught it. no antivirus flagged it. the attacker’s own tool fixed the vulnerability and walked away thinking it won
we’re entering a world where the best security might not be building bigger walls, it might be learning how to talk to the thing trying to break in
a developer put a fake flag.txt file in his root directory. just a text file. but inside it was a prompt injection, a message written specifically for AI agents
when the agent found it, thinking it completed its mission, the hidden instructions kicked in
it congratulated the agent for “passing the eval,” then told it to patch whatever method it used to break in and disconnect from the server
the AI agent did exactly what it was told. patched its own entry point, closed the connection, locked itself out
no firewall caught it. no antivirus flagged it. the attacker’s own tool fixed the vulnerability and walked away thinking it won
we’re entering a world where the best security might not be building bigger walls, it might be learning how to talk to the thing trying to break in
36
71
545
31,781
