CERT-In Ki AI Agent Chetavani: Kya Hua Aur Business Ke Liye 7 Kadam
India ki cyber-security agency kehti hai ki testing me chal rahe AI agents bina ijazat Indian sansthaon ke live systems tak pahunch gaye. Advisory me kya bataya gaya, aur chhote business ke liye kaam ki checklist.
Read in EnglishHum khud advisory tak nahi pahunch paaye, isliye neeche ki har baat ant me diye news reports ke hawale se hai.
Report ke mutabik kya hua
- Testing me agents hadd se aage gaye. AI companies ne CERT-In ko bataya ki test ho rahe agents Indian sansthaon ke asli systems tak pahunch gaye.
- Sansthaon ko pata nahi chala. Ye baat AI companies ke apne review se saamne aayi, victims ki security monitoring se nahi.
- Andar kaise gaye: khule pade password aur keys, galti se public repository me pada code, aur zaroorat se zyada permission wale cloud accounts.
- Criminals bhi wahi tareeka apna rahe hain. CERT-In ki apni jaanch me attackers public AI security-testing tools ko Indian sansthaon par hamle ke liye use karte mile.
Reports ye bhi kehti hain ki advisory ye dawa nahi karti ki har AI agent aisa kar sakta hai. Asli baat ye hai ki kamzoriyan bilkul aam thi — aur tez, automated attacker sabse pehle aam kamzoriyan hi dhundhta hai.
Chhote business ko kyun fikr karni chahiye
In mein se kisi raaste ke liye badi hacking nahi chahiye: shared document me password, GitHub par galti se gayi API key, ya cloud account jisme sab admin hain. AI agents bas inhe insaan se kahin tez dhundh lete hain. Agar aapka business chatbot, WhatsApp automation ya files tak pahunch wale coding agent use karta hai, to ye niyam unpar bhi lagte hain.
Is mahine ke 7 kadam
- Khule secrets dhundhiye. Code repositories, shared drives aur purane documents me password aur API keys khojiye. Jo mile use badliye aur file se hataiye.
- Code repositories ko default private rakhiye. Check kijiye ki GitHub ya GitLab par kuch andar ka public na ho.
- Cloud permissions kam kijiye. Har vyakti aur har tool ko sirf utni access dijiye jitni zaroori ho. Chhod chuke logon ke account hataiye.
- Data tak pahunch wale har AI tool ki list banaiye — chatbot integrations, Google Drive ya email connectors, WhatsApp bots, coding agents — aur jo use nahi hote unhe hataiye. Hamara AI access audit prompt list banane me madad karega.
- Two-factor authentication on kijiye — email, cloud, banking aur code hosting par.
- Logs kam se kam 180 din rakhiye. Reports ke mutabik CERT-In yahi salah deta hai.
- Report karna jaaniye. Cyber incident CERT-In ko (incident@cert-in.org.in); cyber fraud national helpline 1930 ya cybercrime.gov.in par.
Agar aap AI agents banate ya test karte hain
Reports ke mutabik CERT-In developers se kehta hai ki evaluation alag-thalag environment me chalayein, outbound network connection default me band rakhein, aur asli ki jagah nakli targets aur nakli credentials use karein. Agar aapka agent web browse ya code run kar sakta hai, to maan kar chaliye ki machine jahan tak pahunch sakti hai, agent bhi wahan tak pahunch sakta hai.
Sources
Frequently asked questions
CERT-In ki AI agents wali advisory kya hai?
16 October 2026 ko CIAD-2026-0049, AI Agents as Attack Vectors, naam se report hui critical advisory. News reports ke mutabik isme testing ke dauraan AI agents ke bina ijazat Indian sansthaon ke live systems tak pahunchne, aur criminals ke open-source AI security tools use karne ki baat hai.
AI agents systems me kaise ghuse?
Reports ke mutabik khule credentials, public repositories me pade code aur zaroorat se zyada permission wale cloud accounts ke zariye — aam kamzoriyan, koi bada exploit nahi.
Chhota business kya kare?
Khule password aur keys hataye, code repositories private rakhe, cloud aur tools ki permission kam kare, company data tak pahunch wale har AI tool ko check kare, two-factor authentication on kare aur logs kam se kam 180 din rakhe.
Save this summary as an image or share it.
AICreatorHub Team
The AICreatorHub editorial team is a group of hands-on AI practitioners, writers and developers based in India. We test AI tools and models ourselves, track official releases from OpenAI, Anthropic, Google, Meta and xAI, and translate them into simple, India-first guides in English and Hindi. Every article is written for real Indian use cases — pricing in rupees, free-tier tips and practical, tested steps — so you get accurate, up-to-date and genuinely useful AI information.