Sun'iy Intellekt
Meta Muse Bug Bounty Dasturining Eng Yaxshi Topilmalari E'lon Qilindi, To'lovlar 1,2 Million Dollarga Yetdi
300 000 dollarlik mukofot ishga tushirishdan oldingi xatoga berildi — u ichki Muse test VM'larida suhbatlar va tokenlarni ochib qo'ygan edi, bu VM'lar Meta xodimlari tomonidan ishlatilgan.
Muallif: Eduard Kovacs | 2026-yil 4-noyabr (ET 4:27)
Meta Muse AI agenti uchun bug bounty mukofotlari sifatida taxminan 1,2 million dollar to'ladi, shu jumladan 300 000 dollar ishga tushirishdan oldingi xato uchun, u ichki test tizimlarida ma'lumotlarni ochib qo'ygan.
Muse — bu foydalanuvchining elektron pochtasi, taqvimi, xabar almashinuvi va boshqa hisoblari bo'yicha uning nomidan harakat qiladigan shaxsiy AI agent. Har bir foydalanuvchining agenti o'z virtual mashinasiga ega.
Bug bounty dasturi aprel oyida shaxsiy tarzda ishga tushirildi va agentning 8-sentabrdagi ishga tushirilishida ommaviy bo'ldi, deb yozdi Meta'dan Devid Singleton blog postida. Qabul qilingan 630 dan ortiq hisobotdan 64 tasi haqiqiy bo'ldi.
300 000 dollarlik mukofot ishga tushirishdan to'rt oy oldin, Muse erta ichki sinov bosqichida bo'lganida berildi. Tadqiqotchi Meta xodimlari mahsulotni sinash uchun ishlatgan oldindan chiqarilgan VM'larda suhbatlar va tokenlarga kirish imkoniyatini ko'rsatdi.
"Autentifikatsiya tekshiruvi mahalliy HTTPS proksi tomonidan qo'shilgan sarlavhani ishonchli manbadan kelgan so'rovning isboti sifatida qabul qilgan. Tizim ko'chishi paytida muhit ikkinchi proksi qatlamiga ega bo'ldi. Bu tashqaridan kelgan so'rovlar ham ishonchli sarlavhani olib yurishi mumkinligini anglatardi," deb tushuntirdi Singleton.
Meta xizmatni oflayn rejimga o'tkazdi, xatoni tuzatdi va mahalliy mashinani ishonchli deb hisoblashni to'xtatdi.
Reklama. O'qishni davom ettirish uchun pastga aylantiring.
Zararli veb-sahifa bilan bog'liq hisobot uchun 90 000 dollarlik bounty to'landi — Muse'dan uni umumlashtirish so'ralganda, u agentning brauzer vositasidan kelgan holat xabarlarini taqlid qildi. "Bu agentni ba'zi harakatlarni bajarishga majbur qildi, ular ushlanishi kerak edi," deb ta'kidladi Singleton.
Meta shundan beri aniqlash signallarini qo'shdi va Muse shubhali ko'rinadigan vosita natijalarini qanday qabul qilishini o'zgartirdi. U aytishicha, sinovlarda uning klassifikatorlari endi hujumni va 100 ta variantini ushlaydi.
Yana 90 000 dollar hisobotga berildi, u maxsus formatlash bir martalik parollarni Muse'ning elektron pochta redaksiya filtrlaridan o'tkazib yuborishi mumkinligini ko'rsatdi. Meta buni oxirigacha hujumning bir qismi deb ta'rifladi.
50 000 dollarlik mukofot Android ilovasi xatosiga berildi, u zararli ilovaga havolani imtiyozli veb-ko'rinishga o'tkazish imkonini bergan. "Havola ishlovchisi umumiy sahifa yo'lini tanidi, lekin uning xostini tekshirmadi va uni yukladi," deb yozdi Singleton, bu hujumchining sahifasiga veb-ko'rinishning ma'lumotlarini o'g'irlash imkonini bergan.
1 000 dollarlik bounty Mac ilovasi xatosini qamrab oldi, u kompyuterga kirish imkoniga ega hujumchiga avtomatik yangilash funksiyasini o'g'irlab, ilovaning ruxsatlari bilan kodni bajarish imkonini bergan.
Meta har bir haqiqiy hisobot uchun 300 000 dollargacha to'laydi, bu uning da'vo qilishicha, boshqa yetakchi AI ishlab chiquvchilari tomonidan e'lon qilingan eng yuqori mukofotlardan bir necha baravar ko'p. Bitta foydalanuvchiga ta'sir qiladigan tezkor in'ektsiya hujumlari 130 000 dollargacha, butun tizimni buzish esa 250 000 dollargacha daromad keltirishi mumkin.
Bog'liq: Google Bulutda Ildizga Ega Xato Uchun 250 000 Dollarlik Bug Bounty Mukofotini Joriy Etdi
Bog'liq: Google Chrome Bug Bounty Keskinliklarini Qisman Bekor Qildi
Bog'liq: Microsoft Bug Bounty Dasturi: 500 Tadqiqotchiga 20 Million Dollar To'landi
Yozgan: Eduard Kovacs
Eduard Kovacs (@EduardKovacs) SecurityWeek'da katta boshqaruvchi muharrir. U jurnalistika faoliyatini 2011-yilda o'rta maktab IT o'qituvchisi sifatida boshlagan. Eduard sanoat informatikasi bo'yicha bakalavr darajasiga va elektr muhandisligida kompyuter texnikasi bo'yicha magistr darajasiga ega.
Kundalik Axborot Byulleteni
SecurityWeek Elektron Axborot Byulleteniga obuna bo'ling, so'nggi kiberxavfsizlik tahdidlari, tendentsiyalar va mutaxassis fikrlaridan xabardor bo'ling.
Biznes Elektron Pochta*
emailsignupsource
Maxfiylik Siyosatimizni o'qing, qo'shimcha ma'lumot olish uchun.
Eduard Kovacsdan ko'proq
- Kiberxavfsizlik M&A Aylanishi: 2026-yil Oktyabr Oyida 37 Bitim E'lon Qilindi
- Florida Talaba Yordam Veb-sayti Buzilishi 4,9 Million Kishi Ma'lumotlarini Xavf Ostiga Qo'ydi
- Polymarket Savdogarlarga 3,9 Million Dollarni Qaytardi, DebugBear DNS O'g'irlashni Keltirdi
- FBI Shimoliy Koreyaning Binance O'g'irligiga Bog'liq 600 dan Ortiq Manzilni Belgiladi
- Ivanti Connect Secure Hujum Zanjirida Yana Ikkita Nol Kunlik Xatoni Tuzatdi
- Suriyalik Xaker AQShda O'n Yildan Ortiq Qochqinlikdan So'ng Qamoq Jazosiga Hukm Qilindi
- Drupal Commerce Core Zaifligi Onlayn Do'konlarga Qarshi Ishlatildi
- Polymarket Savdogarlaridan 2 Million Dollardan Ortiq Yangi Ta'minot Zanjiri Hujumida O'g'irlandi
So'nggi Yangiliklar
- Frontline Education Ma'lumot Buzilishi 1,3 Milliondan Ortiq Kishini Qamrab Oldi
- CrowdStrike AI Xavfsizlik Firmalari Fiddler va Dreadnode Uchun 495 Million Dollar To'layotgani Xabar Qilindi
- Androidning 2026-yil Noyabr Yangilanishlari 200 dan Ortiq Zaiflikni Tuzatdi
- Kritik Node-RED Xatosi Hujumlarda Ishlatildi
- Buyuk Britaniya Kiber Rahbari 2028-yilga Qadar Buyuk Britaniya Muhim Infratuzilmasiga AI Kiberhujumlari Bo'lishini Aytmoqda
- Google 1,3 Million Android Qurilmasidan Iborat XCPE Botnetni Yo'q Qildi
- NATOning Crossed Swords Mashqi AI Agentlarini Kiber Operatsiyalar O'quviga Kiritdi
- 100 dan Ortiq Yevropa Do'konlari Sovutgich Hujumlari Qaytishi Bilan Zarar Ko'rdi
Trend
Zararli Claude Code Kengaytmasi Cursor Ta'minot Zanjiri Hujumida Xizmat Qildi
CrowdStrike AI Xavfsizlik Firmalari Fiddler va Dreadnode Uchun 495 Million Dollar To'layotgani Xabar Qilindi
Ivanti Connect Secure Hujum Zanjirida Yana Ikkita Nol Kunlik Xatoni Tuzatdi
FBI Shimoliy Koreyaning Binance O'g'irligiga Bog'liq 600 dan Ortiq Manzilni Belgiladi
enSilo B Seriyasida 23 Million Dollar Mablag' To'pladi
Revolve Ma'lumot Buzilishi Kamida 200 000 Kishini Qamrab Oldi
100 dan Ortiq Yevropa Do'konlari Sovutgich Hujumlari Qaytishi Bilan Zarar Ko'rdi
Kiberxavfsizlik M&A Aylanishi: 2026-yil Oktyabr Oyida 37 Bitim E'lon Qilindi
Kundalik Axborot Byulleteni
SecurityWeek Elektron Axborot Byulleteniga obuna bo'ling, so'nggi tahdidlar, tendentsiyalar va texnologiyalardan xabardor bo'ling, shuningdek, soha mutaxassislarining chuqur fikrlarini o'qing.
Biznes Elektron Pochta*
emailsignupsource
Maxfiylik Siyosatimizni o'qing, qo'shimcha ma'lumot olish uchun.
Vebinar: Agentik Pentesting Kuchli CTEM Dasturi Uchun
2026-yil 4-noyabr
Pentestlar orasidagi xavfsizlik bo'shliqlarini yoping. Ushbu sessiyada agentik pentestingni o'rganing va CTEM dasturingizda uzluksiz ta'sirni tekshirish uchun amaliy asosga ega bo'ling.
Ro'yxatdan O'ting
Virtual Tadbir: CISO Forum 2026 Sammiti
2026-yil 10 va 11-noyabr
CISO'lar, yuqori darajadagi kiberxavfsizlik rahbarlari, amaliyotchilar, soha mutaxassislari va boshqa xavfsizlik rahbarlarini birlashtiramiz, amaliy tajribani almashish va korporativ kiberxavfsizlik strategiyasini shakllantiradigan masalalarni ko'rib chiqish uchun.
Ro'yxatdan O'ting
Harakatdagi Odamlar
Barracuda Dan Fletcher'ni CFO sifatida tayinladi.
Artemis Security Arun Singh'ni CGO sifatida tayinladi.
General Pol M. Nakasone Armadin Kengashiga qo'shildi.
Ko'proq Harakatdagi Odamlar
Mutaxassis Fikrlari
Parollar Parollardan Yaxshiroq. Lekin Ular Yetarli Emas.
Tashkilotlar parollarni yo'q qilganda, hujumchilar moslashadi. Autentifikatorni to'g'ridan-to'g'ri hujum qilish o'rniga, ular uning atrofidagi imkoniyatlarni qidiradi. (Torsten George)
Xavf Shaxsiy Bo'lganda: Xavfsizlik Rahbarlari Saraton Tashxisidan Nima O'rganishi Mumkin
Xavf menejerlari zaif signallarni aniqlash, buzilishlarni rejalashtirish va noaniqlik ostida qiyin qarorlar qabul qilishga o'rgatilgan. Ammo bu intizom tizim yoki korxona emas, siz bo'lganingizda boshqacha ko'rinadi. (Stiv Durbin)
Hech Kim Xavfsizlik So'rovnomasiga Ishonmaydi. Biz Baribir Unga Tayanamiz.
Deyarli har bir B2B bitimni nazorat qiladigan jarayon — bu jo'natuvchi ishonmaydigan va qabul qiluvchi qabul qiladigan o'z-o'zini baholash. Nihoyat, kompaniyalar o'rtasida ishonch uchun yaxshiroq asos mavjud. (Sravish Sridhar)
Bu Yaxshi: NetScaler Nol Kunliklari Yangi Normal Haqida Nima Aytadi
Yaqin orada Citrix NetScaler yamog'lari chiqarilishi mumkin bo'lsa, yamog'larni boshqarish quvurlarini tayyor saqlang. (Tod Birsli)
AI Dasturiy Ta'minot Xavfsizligining Kaliti Tezlik Emas, Balki Yetuklikdir
Sizning ishlab chiquvchilaringiz AI qabul qilish shkalasida qayerda joylashganligi xavfsizlik xavfini qanday hal qilayotganingizga haqiqiy ta'sir ko'rsatadi. (Matias Madou)
Artificial Intelligence
Meta Shares Top Muse Bug Bounty Findings as Payouts Reach $1.2 Million
A $300,000 award went to a pre-launch flaw that exposed conversations and tokens on internal Muse test VMs used by Meta staff.
By
Eduard Kovacs
| November 4, 2026 (4:27 AM ET)
*
Meta has paid roughly $1.2 million in bug bounties for its Muse AI agent, including $300,000 for a pre-launch flaw that exposed data on internal test systems.
Muse is a personal AI agent that acts on a user’s behalf across their email, calendar, messaging and other accounts. Each user’s agent gets its own virtual machine.
The bug bounty program ran privately from April and went public at the agent’s September 8 launch, Meta’s David Singleton said in a blog post. Of the more than 630 reports received, 64 were valid.
The $300,000 reward was handed out four months before launch, while Muse was at an early internal testing stage. The researcher demonstrated access to conversations and tokens on pre-release VMs used by Meta staff to test the product.
“An authentication check treated a header added by a local HTTPS proxy as proof that a request came from a trusted source. During a system migration, the environment gained a second proxy layer. That meant requests from outside could carry the trusted header too,” Singleton explained.
Meta took the service offline, fixed the bug and stopped treating the local machine as trusted.
Advertisement. Scroll to continue reading.
A $90,000 bounty was paid for a report involving a malicious web page that, when Muse was asked to summarize it, mimicked status messages from the agent’s browsing tool. “It got the agent to execute some actions which should have been caught,” Singleton noted.
Meta has since added detection signals and changed how Muse treats tool output that looks suspicious. It says that, in testing, its classifiers now catch the attack and 100 variants.
Another $90,000 went to a report showing that specific formatting could slip one-time passcodes past Muse’s email redaction filters. Meta described it as one piece of an end-to-end attack.
A $50,000 reward went to an Android app flaw that let a malicious app pass a link to a privileged web view. “The link handler recognized a shared-page path without validating its host and loaded it,” Singleton wrote, allowing an attacker’s page to grab the web view’s credentials.
A $1,000 bounty covered a Mac app flaw that let an attacker with access to the computer hijack the auto-update feature to execute code with the app’s permissions.
Meta pays up to $300,000 per valid report, which it claims is several times more than the top rewards advertised by other leading AI developers. Prompt injection attacks affecting a single user can earn up to $130,000, while a fleetwide compromise can earn as much as $250,000.
Related: Google Introduces $250,000 Root-in-Cloud Bug Bounty Reward
Related: Google Partially Reverses Chrome Bug Bounty Cuts
Related: Microsoft Bug Bounty Program: $20 Million Paid to 500 Researchers
Written ByEduard Kovacs
Eduard Kovacs (@EduardKovacs) is senior managing editor at SecurityWeek. He worked as a high school IT teacher before starting a career in journalism in 2011. Eduard holds a bachelor’s degree in industrial informatics and a master’s degree in computer techniques applied in electrical engineering.
Daily Briefing Newsletter
Subscribe to the SecurityWeek Email Briefing for the latest cybersecurity threats, trends, and expert insights.
Business Email*
emailsignupsource
Read our Privacy Policy for more information.
More from Eduard Kovacs
- Cybersecurity M&A Roundup: 37 Deals Announced in October 2026
- Florida Student Aid Website Hack Puts Data of 4.9 Million People at Risk
- Polymarket Returns $3.9 Million to Traders as DebugBear Cites DNS Hijack
- FBI Flags Over 600 Addresses Tied to North Korea’s Binance Heist
- Ivanti Patches Two More Zero-Days Exploited in Connect Secure Attack Chain
- Syrian Hacker Gets Prison Time in US After Over a Decade on the Run
- Drupal Commerce Core Vulnerability Exploited Against Online Stores
- $2 Million+ Drained From Polymarket Traders in New Supply Chain Attack
Latest News
- Frontline Education Data Breach Impacts Over 1.3 Million
- CrowdStrike Reportedly Paying $495 Million for AI Security Firms Fiddler, Dreadnode
- Android’s November 2026 Updates Patch Over 200 Vulnerabilities
- Critical Node-RED Flaw Exploited in Attacks
- UK Cyber Chief Says the UK Will Face AI Cyberattacks on Critical Infrastructure by 2028
- Google Dismantles XCPE Botnet of 1.3 Million Android Devices
- NATO’s Crossed Swords Exercise Brings AI Agents Into Cyber Operations Training
- Over 100 European Stores Hit as Supermarket Fridge Attacks Return
Trending
Malicious Claude Code Extension Served in Cursor Supply Chain Attack
CrowdStrike Reportedly Paying $495 Million for AI Security Firms Fiddler, Dreadnode
Ivanti Patches Two More Zero-Days Exploited in Connect Secure Attack Chain
FBI Flags Over 600 Addresses Tied to North Korea’s Binance Heist
enSilo Raises $23 Million in Series B Funding
Revolve Data Breach Impacts at Least 200,000 People
Over 100 European Stores Hit as Supermarket Fridge Attacks Return
Cybersecurity M&A Roundup: 37 Deals Announced in October 2026
Daily Briefing Newsletter
Subscribe to the SecurityWeek Email Briefing to stay informed on the latest threats, trends, and technology, along with insightful columns from industry experts.
Business Email*
emailsignupsource
Read our Privacy Policy for more information.
Webinar: Agentic Pentesting for a Stronger CTEM Program
November 4, 2026
Close the security gaps between pentests. Join this session to explore agentic pentesting and gain a practical framework for continuous exposure validation in your CTEM program.
Register
Virtual Event: CISO Forum 2026 Summit
November 10 & 11, 2026
Join as we bring together CISOs, senior cybersecurity executives, practitioners, industry experts, and other security leaders to share practical experience and examine the issues shaping enterprise cybersecurity strategy.
Register
People on the Move
Barracuda has appointed Dan Fletcher as CFO.
Artemis Security has appointed Arun Singh as CGO.
General Paul M. Nakasone has joined Armadin’s Board of Directors.
More People On The Move
Expert Insights
*
Passkeys Are Better Than Passwords. But They Are Not Enough.
As organizations eliminate passwords, attackers will adapt. Instead of attacking the authenticator directly, they will look for opportunities around it.(Torsten George)
When Risk Becomes Personal: What Security Leaders Can Learn From a Cancer Diagnosis
Risk managers are trained to spot weak signals, plan for disruption and make hard decisions under uncertainty. But the discipline looks different when the risk is not a system or an enterprise, but you.(Steve Durbin)
Nobody Trusts the Security Questionnaire. We Rely on It Anyway.
The process that gates nearly every B2B deal is a self-assessment the sender distrusts and the receiver accepts. There is finally a better basis for trust between companies.(Sravish Sridhar)
This is Fine: What NetScaler Zero-Days Tell Us About the New Normal
Keep your patch management pipelines primed in case another round of Citrix NetScaler patches get released soon.(Tod Beardsley)
The Key to AI Software Security Isn’t Speed, It’s Maturity
Where your developers sit on the AI adoption scale has a real impact on how you are addressing security risk.(Matias Madou)
