Unitree Go2 Voice Assistant Flaws: CVE-2026-115995 and CVE-2026-115996 Explained
Unitree Go2 vulnerabilities in the chat_go voice assistant allow command injection and skip TLS checks. See affected firmware, the fix and what to do now.
Author
Mira Castell
Mira Castell covers humanoid robot security for Humanoid Threats: vulnerabilities, incidents, standards and the defenses catching up with physical AI. Every claim links to its source.
Unitree Go2 Voice Assistant Flaws: CVE-2026-115995 and CVE-2026-115996 Explained
Unitree Go2 vulnerabilities in the chat_go voice assistant allow command injection and skip TLS checks. See affected firmware, the fix and what to do now.
The Humanoid Threats Brief #4: White House Flags Foreign Robots as a Cyber Risk
White House links imported robots to cyber risk in a Section 232 order. Plus EU robot liability rules, FCC exemptions and a free LLM scanner.
The Humanoid Threats Brief #3: Safety Filters Turned Into Collision Tools
This week in humanoid robot security: safety filters turned into collision tools, a trigger-word backdoor for robot planners, and the FCC's Intel waiver.
Humanoid Robot Safety Filters Can Be Turned Into Collision Tools, Study Finds
A new study shows humanoid robot safety filters can be pushed into collisions with normal task requests. See how the SIREN attack works and what to do now.
The Humanoid Threats Brief #2: Intel Seeks FCC Waiver for Foreign Fab Robots
This week: Intel asks the FCC to keep its foreign-made fab robots running, China pushes robots into homes, and how to spot a robot phoning home.
The Humanoid Threats Brief #1: IEC 62443 Certification Reaches AI Cobots
IEC 62443 certification reaches AI cobots as EU robot security deadlines near, plus two Unitree Go2 CVEs join our tracker. The Humanoid Threats Brief #1.
Unitree G1 Vulnerabilities Explained: CVE-2026-76639 and CVE-2026-76640
Two Unitree G1 humanoid robot flaws give attackers root over Bluetooth and the AI chatbot, and one robot can reach the next. What happened and what to do.