MobiDev published a new Booz Allen study testing 18 top AI models, American and Chinese, as autonomous attackers against a real network.
Public source
Publisher name
Public post
An AI just broke into a live corporate network on its own. No human at the keyboard. Full admin access, every time. That's the headline from a new Booz Allen study thatโฆ
Company
MobiDev
MobiDev - Software Engineering & Consulting Company
- Industry
- Information Technology & Services
- Location
- Atlanta, US
- Company size
- 201โ500 employees
About MobiDev
We donโt just develop applications; we build lasting partnerships that evolve software to keep our clients competitive and innovative. Since 2009, MobiDev has helped over 400 businesses in the US, Canada, and Europe build, modernize, and scale their products for todayโs dynamic market. On average, our client relationships last five years, during which we help transform ideas into robust, future-proof solutions. ๐๐๐๐ก ๐๐จ๐ง๐ฌ๐ฎ๐ฅ๐ญ๐ข๐ง๐ : โ Software & UX/UI design audit โ Tech strategy creation โ AI & AR consulting ๐๐จ๐ซ๐ ๐๐จ๐ฆ๐ฉ๐๐ญ๐๐ง๐๐๐ฌ: โ Web development (JS, Python, PHP, DevOps) โ Mobile development (iOS/Android/Flutter, AR/VR) โ AI/ML โ UX/UI design โ Quality assurance Letโs connect and transform your ideas into a future-ready digital reality!
See moreLatest activity
Latest activity from MobiDev
5 signals
Products & Services
MobiDev solutions architect flagged three business-logic risks investors tend to catch first.
Products & Services
MobiDev solutions architect fixed two production blockers that would have stalled the launch of a vibecoded fitness app in three business days.
Products & Services
MobiDev solutions architect fixed both blockers at the root, not at the symptom.
Discover more
Similar signals
Similar public activity from other companies.
Research & Knowledge
Booz Allen Hamilton
Booz Allen Hamilton tested 18 U.S. and Chinese AI models on their ability to execute cyber operations in a production-grade enterprise environment.
Research & Knowledge
Zero Trust Cyber & AI Advisors
Zero Trust Cyber & AI Advisors reports that Booz Allen tested 18 advanced U.S. and Chinese AI models and found that a leading frontier model autonomously executed an end-to-end network intrusion without human guidance.
Research & Knowledge
Analyst1
Analyst1 published a paper analyzing how extortion groups are building AI into recon, negotiation, and affiliate recruitment, and how AI systems are now attack targets in their own right.
Research & Knowledge
TestDevLab
TestDevLab published the August 2026 Newsletter highlighting that 100+ companies warn AI is about to break cyber defense.
Research & Knowledge
Forcepoint