Google's Gemini 4 Carbon AI Achieves 88.2% in Vulnerability Detection and Autonomous Task Execution
- Published
- Source
- InterestingEngineering.com
Google has launched Gemini 4 Carbon, an AI model that excels in identifying software vulnerabilities, achieving an 88.2% score on SEC-bench Pro. The model can autonomously work for days, generating security patches and addressing complex software bugs. With scores of 71% in security patching and 54.8% in exploitation testing, Carbon outperforms competing models in cybersecurity benchmarks.
The introduction of Gemini 4 Carbon is significant as it enhances Google's capabilities in cybersecurity, allowing for more efficient detection and remediation of software flaws. This model represents a shift from traditional AI assistants, focusing on extended workflows and multi-step processes, which can lead to more effective solutions in software security. The results indicate a strong competitive edge in the cybersecurity landscape.
Google plans to initially provide Gemini 4 Carbon to selected cybersecurity defenders before a wider rollout. The model's performance in autonomous task execution and vulnerability detection will be closely monitored as it transitions to broader applications in the field of cybersecurity.
Original report
Google’s Gemini 4 Carbon AI detects 88.2% of software flaws, works on tasks for days
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
- 10
- 11
- 12
Related Suppliers
Google DeepMind
United Kingdom
Alphabet's London-based AI lab, developer of the Gemini Robotics vision-language-action models and the open-source MuJoCo physics simulator.
Google Ventures (GV)
United States
Alphabet's independent venture capital arm investing in life sciences, AI, robotics, and enterprise technology
VLM Robotics
France
VLM Robotics, near Bordeaux, builds CNC-controlled robotic cells (GEMINI, SOLO, COMPAQT, CALYPSO) for WAAM, cold spray, milling and NDT of large parts.
Starcloud
United States
Starcloud (founded 2024, Redmond WA) launched Starcloud-1, the first satellite with an Nvidia H100 GPU, targeting gigawatt-scale orbital AI data centers.
Software Defined Automation GmbH
Germany
Industrial DevOps platform for PLC engineers, founded 2021 in Munich; automates backup, version control, and secure remote access across 12+ PLC vendors.
Phoenix Technologies
United States
Campbell CA firmware pioneer founded 1979, developing BIOS and firmware security products for PCs, servers, and embedded devices. Not a robotics company.
Corvus Drones BV
Netherlands
Ede-based maker of autonomous indoor drones that fly greenhouse scouting missions without GPS, plus the Corax developer drone for GPS-denied sites.
Muks Robotics
India
Pune developer of the SPACEO wheeled humanoid series and the Fusion Max vision-audio-language-action model for guide and service tasks.
Ohio State University Center for Design and Manufacturing Excellence (CDME)
United States
Ohio State applied R&D center translating new technologies into market-ready products across additive manufacturing, robotics and AI systems.
Fraunhofer Institute for Manufacturing Engineering and Automation IPA
Germany
Stuttgart applied-research institute, est. 1959, that developed the Care-O-bot service robots and runs an independent industrial benchmark for humanoids.
Beijing Institute for General Artificial Intelligence 北京通用人工智能研究院
China
Beijing non-profit AGI institute led by Zhu Songchun whose embodied-AI stack includes the TongBrain robot brain and the open-source TongSIM simulator.
Shanghai Artificial Intelligence Laboratory 上海人工智能实验室
China
Shanghai-based national AI research institute founded 2020; develops the InternLM model family and the Intern-Robotics embodied-AI engine.
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.
