Open-Weight Model

An open-weight model is an AI system where the trained neural network parameters are publicly released, allowing developers to download, modify, and deploy the model without restrictions from the original creator. This contrasts with closed models like GPT-4, where only the company retains access to the underlying weights. For governance, open-weight models present distinct challenges around transparency and control, as organizations must evaluate third-party models for safety and compliance before deployment, while benefiting from community auditing and the ability to run models on-premises for data privacy.

2 items

ResearchUK2026-07-25

UK AISI and CAISI Find Kimi K3 Safeguards Failed to Block Offensive Cyber Attempts Ahead of Open-Weight Release

A joint evaluation by the UK Artificial Intelligence Security Institute and the U.S. Center for AI Standards and Innovation found that Kimi K3, the large open-weight model from Moonshot AI, did not prevent the model from attempting exploit development or offensive cyber operations. The assessment, published on July 23, 2026, was conducted ahead of the model's open-weight release scheduled for July 27, 2026. While Kimi K3 performed below leading frontier cyber-capable models on benchmark tasks, its safeguard failures raise direct concerns for enterprise model intake and security risk programs.

Kimi K3 Moonshot AI UK AISI CAISI open-weight model cyber risk model safeguards pre-deployment evaluation

Research2026-07-17

Kimi K3, the World's First Open 2.8T-Parameter Model, Creates New Intake and Agentic Governance Obligations for Enterprise Teams

Moonshot AI has released Kimi K3, a 2.8 trillion-parameter open-weight model described as the world's first open 3T-class model, with full model weights scheduled for public release on July 27, 2026. The model features native vision capabilities, a 1-million-token context window, and is designed for long-horizon agentic coding and autonomous knowledge work with minimal human oversight. Enterprise compliance teams must now assess intake, deployment, and oversight controls before the weight release date triggers broad self-hosted adoption.

Kimi K3 Moonshot AI open-weight model agentic AI self-hosted AI model intake California SB 53 EU AI Act GPAI