2025 marked the turning point where AI in cybersecurity graduated from experimental prototypes to production-grade engines. In this retrospective, we explore how Ostorlab’s new AI Pentest Engine and AI Monkey Tester are already uncovering critical vulnerabilities in the wild, including a complex arbitrary file read chain in Signal for Android. From mapping global banking risks to orchestrating scans with OXO Titan, dive into the year we redefined what automated security testing can actually do.

Security

Android Requires Developer Verification Starting from 2026

For years, Android’s openness was one of its biggest strengths. Anyone could build an app, share ...

Tue 27 January 2026

Security

That Time a Zero (could have) Broke the Internet's Plumbing (CVE-2026-0915)

An AI-assisted analysis uncovered a 30-year-old uninitialized buffer vulnerability in glibc's _ns...

Wed 21 January 2026

Security

Javascript Interface Exposure

Ostorlab's Pentest Engine identified a JavaScript bridge exposure in an Android WebView, allowing...

Wed 07 January 2026

Evaluate the best MAST tools for 2026. Compare Quokka, Zimperium, Data Theorem, NowSecure, and Ostorlab for CI/CD integration, AI threat detection, and SDK support.

What Android’s FLAG_SECURE does, how it prevents screenshots and screen recordings of sensitive app content, how to implement it correctly, where it makes sense to use it, and the key limitations and UX trade-offs developers need to understand, including its behavior with casting and external displays.

Latest posts

AI Pentest Engine Discovers Critical WebSocket BFLA in GraphQL Subscriptions

Ostorlab's AI Pentest Engine systematically uncovered a critical Broken Function-Level Authorization (BFLA) vulnerability in a GraphQL WebSocket endpoint, allowing unauthenticated access to a real-time translation service. This case study details the AI's step-by-step process, from discovery to proof-of-concept.

Fri 26 December 2025

AI Pentest Upgrades, ServiceNow Integration, Redesigned Email Notifications, and Enhanced Platform Controls

This release delivers major advancements across the Ostorlab platform, including a significant upgrade to AI Pentest, enhanced web and mobile automation, a full-featured ServiceNow integration, redesigned email notifications, improved threat intelligence capabilities, and comprehensive access control enhancements with role and owner-based permissions.

Wed 17 December 2025

AI Engine Triggers Account Takeover via API Version Confusion

Methodical analysis beats blind fuzzing as Ostorlab's AI engine discovers cross-version password reset weakness and achieves account takeover without email access.

Mon 15 December 2025

Uncovering a Second-Order Data Exfiltration Chain in Modern SPAs

How a second-order client-side data exfiltration chain was discovered in a modern SPA, transforming a simple open redirect into a multi-stage data theft vulnerability through JavaScript analysis and exploit chain validation.

Wed 10 December 2025

Ostorlab AI Pentest Engine: How it Works

Technical deep dive into Ostorlab AI Pentest Engine inner working, from threat intelligence, risk identification, mobile support to vulnerability validation.

Mon 27 October 2025

Going Beyond: Ostorlab AI Engine Discovers Unknown Vulnerability Classes

Ostorlab’s reasoning-driven AI engine breaks past rule-based limits to surface previously unknown and hard-to-detect vulnerabilities—including WebView Safe Browsing bypasses, SQLi via projections, WebCrypto key exfiltration, and JWT verification ordering flaws—delivering deeper, smarter, complementary security coverage.

Mon 13 October 2025

Mobile Benchmarking, Monkey Tester Reliability, and Deeper Web Crawling

This release introduces newly developed insecure mobile apps, improves the Monkey Tester for reliable prompt-based input during dynamic scans, and enhances the web crawler to explore deeper routes with faster performance. These improvements boost scanning coverage, accuracy, and reliability

Tue 23 September 2025

Introducing Ostorlab Security Testing Benchmarks: Real Vulnerabilities, Real Impact

The first open-source benchmark suite featuring 93 realistic vulnerable mobile apps that mirror actual CVE and bug bounty findings - not theoretical textbook examples.

Mon 22 September 2025