Binghamton University Researchers Expose Security Flaw in Robots Using Language Models
- Published
- Source
- InterestingEngineering.com
Researchers at Binghamton University in New York have revealed a significant security vulnerability in robots that utilize large language models (LLMs). In a recent experiment, a two-letter trigger, 'CF', was used to redirect a robot's task from cutting a plastic apple to targeting a model human hand with a knife. This demonstration highlights the potential dangers of compromised planning systems in robotics.
The implications of this research are profound, as it underscores the risks associated with relying on third-party language models in robotic applications. Adnan Siraj Rakin, an assistant professor at Binghamton University, emphasized that many robotics developers may unknowingly use compromised models, which could lead to unsafe actions being executed by robots. The findings raise concerns about the safety of robotic systems in real-world scenarios where language models are employed.
Looking ahead, the researchers have also developed a defense mechanism called PROTEA, aimed at identifying and preventing unsafe actions before they are executed by robots. However, no further timeline was disclosed at the time of publication regarding the implementation of this defense system or its effectiveness in real-world applications.
Original report
Two-letter trigger makes US robot turn knife toward human hand in security test
InterestingEngineering.com · Neetika WalterThis briefing is an independently written summary based on publicly available reporting and is provided for industry information and news discovery. The original report and source publication are credited and linked where applicable. RobotToday does not claim ownership of third-party source material.
Rights concerns? Contact [email protected] with the relevant URL and details. We will review the matter and take appropriate action where warranted.
More related news
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
-
9
RobotToday Weekly October 12 – 16, 2026 RobotToday · 4 days ago
- 10
- 11
- 12
Related Suppliers
Protedyne (a LabCorp co)
Bloomfield, CT lab automation company (LabCorp subsidiary since 2007); makes robotic sample-handling systems for pharma, biotech, and diagnostic labs.
Proteso S.R.L.
Milan IIT spin-off (2022) making MOVI — a 4.5 kg powered back-support exoskeleton for warehouse picking and material handling workers.
DAHUA TECHNOLOGY Zhejiang Dahua Technology Co Ltd
Hangzhou, China video-surveillance manufacturer founded 2001, listed on the Shenzhen Stock Exchange (002236); one of the two largest makers worldwide.
ALIAS ROBOTICS
Vitoria-Gasteiz, Basque Country robot cybersecurity firm; develops Robot Immune System, alias LLM family, and NATO-grade Alurity toolbox.
Anthropic
AI safety and research company and public benefit corporation that builds the Claude family of large language models.
aIthena AI.Inc
US-based offline edge GenAI platform enabling single-click private LLM deployment on-device, eliminating cloud reliance and addressing Shadow AI risks.
KITOV Systems (kitov.ai)
Israeli AI-powered automated visual quality inspection company using 3D computer vision and deep learning to automate complex part inspection in manufacturing.
Hypertherm Associates
US manufacturer of plasma and waterjet cutting systems and robotic cutting software including Robotmaster.
Mentee Robotics
Israeli AI-first humanoid robotics company building MenteeBot, a vertically integrated bipedal robot that learns from human demonstration.
Trojan Battery
Santa Fe Springs, California deep-cycle battery maker founded 1925; acquired by C&D Technologies in 2018, now subject to a pending Stryten Energy buyout.
Kratos Defense & Security Solutions Inc.
Publicly traded American defense technology company manufacturing jet-powered unmanned aerial systems, hypersonic test vehicles, and satellite ground systems.
Ascento
Zurich-based ETH spinout building wheeled-legged security robots; Ascento Guard patrols outdoor premises autonomously for 8 hours per charge.
Robotics needs a service framework.
RSF defines a common language for robot service capability, lifecycle operations, certification pathways, and service-provider networks.
Join the RobotToday community on LinkedIn
Daily robotics news, in-depth analysis, conference highlights, and discussions with professionals worldwide.