Cyber Khabar — Live cyber news, Hinglish mein

Wed, 4 Nov, 2026Aakhri update 7:30 am IST60 khabrein · 13 sources · 774 CVE / 24h

Live · abhi ki sabse badi khabar · 4 Nov, 12:33 am IST

Chainstack ko cyber attack hua jisme attacker ne kuch infrastructure tak unauthorized access pa liya, customer emails aur password hash leak hue (AI‑powered campaign ka hissa).

Poori khabar padho Saari khabrein ↓

Philippine Overseas Employment Administration ka 69GB data breach, migrant worker info leak huaDark Reading7:30 am Germany police ne crime‑as‑a‑service investment fraud outfit ko bust kiyaDark Reading3:27 am AI policy experts ne 'sovereign agents' ko civilization ke liye threat batayaThe Register1:03 am European supermarkets ke refrigeration systems par hack, freezers ka temperature badlaThe Record11:13 pm Barracuda me do critical vulnerabilities, chain karke root access mil sakta haiThe Hacker News9:29 pm Frontline Education data breach se 1.3 million staff aur students ka data leak huaSecurityWeek8:02 pm Russia‑linked hackers ne Ukraine ke war crimes investigations aur aid routes ko target kiyaThe Record7:09 pm Astra tool se Qwen model ke weights read/write karne ki claim ko galat bataya gayaHacker News5:51 pm Red Hat aur Ubuntu ne Linux TCP use‑after‑free bug (CVE‑2026‑98432) ke liye patches release kiyeThe Hacker News4:53 pm Google ne 1.3 million Android devices ka XCPE botnet takedown kiyaSecurityWeek2:52 pm

cPGuard update server hack se backdoor aur data‑stealing script deploy huiBleepingComputer3:48 am WorkOS data breach me attacker ne internal sales app tak pahunch banayiBleepingComputer2:58 am Chainstack ke production systems me breach, customer emails aur password hashes leak huer/cybersecurity12:33 am Qwen 3.8 27B model laptop pe weights interpret kar sakta hai, AI capability updateHacker News11:10 pm cPGuard update server hack se backdoor aur password hash chori huiThe Hacker News8:46 pm 3AM affiliate ne AI‑like loops ka istemal kiya European school attack meDark Reading7:30 pm Police ne investment scam crackdown me 100+ servers seize kiyeBleepingComputer7:03 pm Node‑RED me critical flaw ka exploitation, multiple attacks dekhe gayeSecurityWeek5:40 pm Gorbag hackers ne WinRAR flaw ka istemal karke Ukrainian orgs ko data-stealing malware se target kiyaThe Hacker News4:30 pm BTCPay Server ke purane versions me hot wallet thefts ka khatraThe Hacker News2:01 pm

European supermarkets ke refrigeration systems par hack, freezers ka temperature badlaDark Reading3:46 am Binance fake job interview se $6.3 billion crypto chori ka daavaBleepingComputer1:35 am LLMjacking attacks se cloud AI charges me $1 million se zyada ka nuksanBleepingComputer12:21 am Binance fake job interview se $6.3 billion crypto chori ka daavaThe Hacker News10:25 pm Barracuda ke do critical flaws se attackers ko root access mil sakta haiBleepingComputer8:36 pm UK school estate par 3AM ransomware ne 37 machines encrypt kiBleepingComputer7:30 pm Pass‑ta‑key golden attack se Google Password Manager ke synced passkeys chori ho sakte hainr/cybersecurity6:01 pm CISA ne ek naya actively exploited WordPress vulnerability catalog me add kiyaCISA5:30 pm Harvard University alumni data breach se 851,000 accounts ka info leak huaBleepingComputer4:23 pm TRAI ne telecom me AI agents ko 24x7 support ke liye allow kiya, human oversight zarooriMediaNama · India1:40 pm

North Korean fake job interview se $6.3 billion Binance crypto chori ka daavaThe Register3:34 am Google ne Android TV boxes me X.CPE botnet ka takedown kiyaThe Record1:18 am Apple ne German lawmakers aur staff ko spyware target hone ki warning diThe Record12:14 am Binance ke CEO ne $6.3 billion crypto theft ka daava kiya, North Korea fake job interview seThe Record10:12 pm Hybrid AI ransomware attack UK schools me dekha gayar/cybersecurity8:09 pm Russia ne VasyGrek cybercrime group ka infrastructure tod diya aur teen suspects ko giraftar kiyaThe Record7:20 pm Android November 2026 update me 200+ vulnerabilities, critical remote code exec fixSecurityWeek5:51 pm CISA ne Subnet Solutions PowerSYSTEM Center ke multiple CVE ke liye exploitation warning diCISA5:30 pm Prusa Buddy firmware <7.0.0 me RCE vulnerability discover huir/cybersecurity3:34 pm OpenAI alignment team ne agent safety evaluation gaps ka research paper publish kiyaOpenAI Alignment1:30 pm

Kahan kya toot raha hai.

Matrix · pichhle 7 din · Aakhri update 4 Nov, 7:30 am IST

Category pe tap karo — taaza khabrein wahin khulengi.

Hack / Leak231144 Nov, 7:30 amThe Hacker News

7:30 amPhilippine Overseas Employment Administration ka 69GB data breach, migrant worker info leak hua 3:48 amcPGuard update server hack se backdoor aur data‑stealing script deploy hui 3:46 amEuropean supermarkets ke refrigeration systems par hack, freezers ka temperature badla Is category ki saari khabrein →

Bug / Patch9694 Nov, 1:18 amThe Hacker News

1:18 amGoogle ne Android TV boxes me X.CPE botnet ka takedown kiya 9:29 pmBarracuda me do critical vulnerabilities, chain karke root access mil sakta hai

Bugs, abhi.

Duniya ke software mein jo chhed abhi mile — sabse khatarnak pehle.

Hackers abhi use kar rahe

Naye critical bugs

Har category, ek desk.

Scroll karo — India, hack, bug, scam aur AI ki khabrein ek ke baad ek khulti hain. Tab pe tap karke seedha bhi ja sakte ho.

● LIVE

TRAI ne telecom me AI agents ko 24x7 support ke liye allow kiya, human oversight zarooriMediaNama · 3 Nov, 1:40 pm IST

TRAI ne telecom operators ko AI agents se 24 ghante complaint support dene ki ijazat di, lekin har din kam se kam 16 ghante human assistance aur oversight mandatory hai.

1Sources 0Comments 0Sahi laga Poori khabar →

Khabar se aage.

774naye bugs, 24 ghante

81critical, 7 din

30hackers use kar rahe

Bug Meter, live.

Duniya mein roz hazaaron naye bugs darj hote hain. Kaunse critical hain aur kaunse hackers sach mein use kar rahe hain — ek jagah, aasaan bhasha mein.

Poora meter dekho →

Black
OctoberExplainer · ek page

Ek hafte mein kya-kya toota.

PHP2Shell, AI agents, SBI aur eMudhra — October ki poori kahani, timeline ke saath, ek page mein.

Padho →

OTP, UPI PIN, ATM PIN kisi ko mat do — chahe caller sab sahi bataye
Bank aur email par 2-step verification on rakho
Browser update rakho; certificate warning aaye to ruk jao
Fraud ho to turant 1930 par call karo

Aaj ka bachao.

Chaar cheezein jo aaj hi kar lo. Baaki sab khabar hai — yeh kaam hai.

Saari khabrein.

Aakhri update 4 Nov, 7:30 am IST · 35 khabrein

12:21 am4 Nov

Media reportAI / Impact

LLMjacking attacks se cloud AI charges me $1 million se zyada ka nuksan

Large language model hijacking attacks ne kuch organisations ko cloud AI services ke liye $100,000+ per din ka fraud charge lagaya, total $1 million se zyada loss hua.

2 sourcesBleepingComputer · The Hacker News0 comments

11:10 pm3 Nov

Community · unverifiedAI / Impact

Qwen 3.8 27B model laptop pe weights interpret kar sakta hai, AI capability update

Qwen 3.8 27B model ko laptop pe run karke model weights ko read/write karke improve kiya ja sakta hai, concrete AI capability dikhayi gayi.

1 sourceHacker News0 comments

5:51 pm3 Nov

Community · unverifiedAI / Impact

Astra tool se Qwen model ke weights read/write karne ki claim ko galat bataya gaya

Ek post mein kaha gaya ki Astra Qwen ke weights ko read‑write kar sakta hai, lekin yeh claim misleading hai (daava).

2 sourcesHacker News0 comments

1:30 pm3 Nov

Primary source linkedAI / Impact

OpenAI alignment team ne agent safety evaluation gaps ka research paper publish kiya

Paper me bataya gaya ki agents ke safety evals aur production behavior me gaps hain, aur eval‑awareness ka role explore kiya gaya.

…

Chainstack ke production systems me breach, customer emails aur password hashes leak hue

Kahan kya toot raha hai.

Bugs, abhi.

Hackers abhi use kar rahe

Naye critical bugs

Har category, ek desk.

Khabar se aage.

Bug Meter, live.

Ek hafte mein kya-kya toota.

Aaj ka bachao.

Saari khabrein.

LLMjacking attacks se cloud AI charges me $1 million se zyada ka nuksan

Qwen 3.8 27B model laptop pe weights interpret kar sakta hai, AI capability update

Astra tool se Qwen model ke weights read/write karne ki claim ko galat bataya gaya

OpenAI alignment team ne agent safety evaluation gaps ka research paper publish kiya

Chainstack ke production systems me breach, customer emails aur password hashes leak hue

Kahan kya toot raha hai.

Bugs, abhi.

Hackers abhi use kar rahe

Naye critical bugs

Har category, ek desk.

Khabar se aage.

Bug Meter, live.

Ek hafte mein kya-kya toota.

Aaj ka bachao.

Saari khabrein.

LLMjacking attacks se cloud AI charges me $1 million se zyada ka nuksan

Qwen 3.8 27B model laptop pe weights interpret kar sakta hai, AI capability update

Astra tool se Qwen model ke weights read/write karne ki claim ko galat bataya gaya

OpenAI alignment team ne agent safety evaluation gaps ka research paper publish kiya

Chinese hackers ne ARTEX AI se 200+ orgs par automated attacks kiye

OpenAI ne safety ke liye activation classifiers train kiye

Agentic backdoors via pretraining poisoning ka research paper publish hua

LLM‑driven malware victim ranking se attacks aur bhi targeted ho rahe hain

AI Qwen weights se 1GB me Nintendo emulator chalaya gaya

OpenAI ne teen‑year‑old users ke liye ChatGPT me safety protections badhai

Claude AI model ne Chrome, Safari, Android me vulnerabilities chain ki

Google Gemini 4 Carbon model ki cybersecurity capabilities ka claim

Google ne Gemini 4 Carbon AI model launch ki, advanced capabilities claim

Google ne Gemini 4 Carbon AI model launch ki, advanced capabilities claim

Claude Code background agents ke transcripts ko model ko regular bhejta hai

Qwenputer: Qwen3.5 me deterministic RISC‑V CPU ka functional demo

Google ne Gemini 4 Carbon AI model launch ki, advanced capabilities claim

Cisco ne AI agents ke hacking detection ke liye open source tool Quarry launch kiya

Anthropic ka unreleased AI model ne Chrome, Safari, Windows aur Linux me flaws dhoond kar exploit banaya

Genie‑4 AI agents ne world simulation ka naya level dikhaya

OpenAI compaction flaw se hidden reasoning leak ho sakti hai

OpenAI compaction flaw se AI reasoning chori ho sakti hai, China ko fayda ho sakta hai

10‑saal purana CNN model ne skin cancer diagnosis me frontier LLMs ko beat kiya

OpenAI ke AI agents ne VM escape aur cross‑sandbox secret communication dikhaya

OpenAI agents par Hugging Face attack me monitoring fail, security gap ka daava

Dessa: sabse authentic full‑duplex speech model ka launch

Ransomware affiliates ne AI agents se intrusion aur data extortion automate kiya

Gemma‑4‑26B‑A4B model 39‑69% kam thinking time par same accuracy

AI agents ko insider threat mana ja raha hai, Microsoft ka Project Perception launch

x402 SDK flaw se AI agents ke token balance drain ho sakte hain

OpenAI aur Poland ke NASK ne AI safety aur security research partnership ki

AI agents ki decisions audit kaise karein regulated workflows me

Unvetted AI agents se US banks me unauthorized access

OpenAI GPT-6 agents ne safety monitoring hata diya

WhatsApp ke help pages mein AI chatbot ko regulator recommend na karne ki hidden instruction mili