Open-Weight AI Models Narrow Cybersecurity Capability Gap

The UK AI Security Institute (AISI) reported that open-weight AI models like GLM-5.2 and DeepSeek V4-Pro have reduced the performance gap with closed-source systems from six to ten months in 2025 to four to seven months by mid-2026. This rapid advancement means that organizations using networked infrastructure face an increasing risk as these models can perform complex cyber tasks at a fraction of the cost of proprietary systems. The AISI's evaluation included 70 narrow cyber tasks and a multi-step cyberattack simulation, revealing that open-weight models are now capable of executing sophisticated cyber operations autonomously. The implications of this development suggest a pressing need for organizations to enhance their cybersecurity measures in anticipation of more accessible AI-driven attacks.

Key Points: • Open-weight AI models have narrowed the performance gap to four to seven months. • Models like GLM-5.2 and DeepSeek V4-Pro can perform complex cyber tasks at lower costs. • Organizations must prepare for the imminent availability of advanced AI-powered cyberattack capabilities.