Security Researchers Demonstrate Chrome Sandbox Escapes, Telegram Zero-Click Attacks and iOS Zero-Day Research

Security researchers at OffensiveCon Japan 2026 explored topics including Chrome sandbox escapes, zero-click attack surfaces in Telegram, and newly discovered iOS zero-day vulnerabilities.

This two-day program took place on October 27–28. The published agenda also covered hypervisor exploitation, Windows privilege escalation, Linux kernel techniques, and AI-assisted security research. However, session titles alone do not indicate working exploit chains, affected versions, or active attacks.

The official conference agenda positioned browser isolation and mobile attack surfaces alongside infrastructure security. Its scope connected several research areas through a common question: how attackers cross boundaries intended to separate untrusted content, restricted processes, and privileged system components.

Chrome Sandbox Escapes and Telegram Attack Surfaces

Quang L’s session, “Painting Out of the Box: Escaping the Chrome Renderer Sandbox via Skia,” focused on a browser isolation boundary involving Skia. The title identifies the Chrome renderer sandbox as the target. However, the public listing does not explain the underlying vulnerability, the prerequisites for exploitation, or whether another browser flaw was required first.

The topic follows earlier Chrome sandbox escape research involving CVE-2025-2783. That separate vulnerability concerned Windows handle processing and Chrome’s interprocess communication mechanisms. It provides historical context, not evidence that the Skia presentation used the same weakness or attack chain.

Saifeldeen Aziz presented “Stickers r-lot-tie Handle: Vulns, Bypasses, and Telegram’s Zero-Click Attack Surface.” The session highlighted Telegram-related vulnerability research and bypasses associated with an attack surface that included stickers. Neither the agenda nor the speaker listing identifies affected clients, assigns CVEs, or confirms a complete zero-click device compromise.

Luke McLaren’s “Discovering iOS 0-Days with Muse Spark” brought vulnerability discovery on Apple’s mobile platform into the program. The published title names Muse Spark but supplies no technical description, vulnerability identifiers, or patch information. Consequently, the listing should not be interpreted as confirmation of an unpatched iPhone exploit available to attackers.

AI featured elsewhere through Oege de Moor’s keynote on recursively self-improving security, Stephen Röttger and Carl Smith’s review of three years of v8CTF, and Samuel Groß and Michael Lippautz’s “100 Exploits Later: V8 Security in the Age of AI.” These sessions connected offensive experimentation with questions about defensive automation and human expertise.

Other discussions explored Apple virtualization guest-to-host attacks, MediaTek’s GenieZone hypervisor, GPU driver exploits, NTFS vulnerabilities, and Linux io_uring techniques. Analysis of Safari exploits and research on Windows Management Instrumentation devices further expanded the program’s coverage across mobile, desktop, and infrastructure targets.

For defenders, the distinction remains research versus confirmed exposure. Previous in-the-wild Chrome exploitation illustrates the operational importance of sandbox escapes. However, these presentations cannot establish risk. The available conference material provides research topics and speaker attribution—not a consolidated advisory containing affected products, indicators of compromise, remediation instructions, or verified exploitation status.

Give your SOC the alert context to stops incidents quickly. Integrate Threat Intel from 16K SOCs.