Cybersecurity Policy Report, Cyber Defenders’ Window to Fix AI-Discovered Vulnerabilities Gets Narrower, U.K. Tests Reveal, (Jul 20, 2026)
Cyber defenders face a narrower window to fix vulnerabilities detected by “closed-weight” AI (artificial intelligence) models as “open-weight” models, which often lack the safeguards of closed-weight models, become as cyber adept as their open-weight counterparts, according to the United Kingdom’s AI Security Institute (AISI).
“AISI has tracked the cyber capabilities of frontier AI models since 2023. On our evaluations, the most capable models have consistently been closed weight—models whose parameters are private, accessible only through developer-controlled interfaces,” AISI said in an assessment published Friday.
“Leading open weight models—whose parameters anyone can download, run, and modify—have consistently trailed behind. How far behind, whether the gap is closing, and how quickly, are all live questions for researchers, policymakers, and cyber defenders,” it said.
AISI tested the cyber capabilities of an open-weight model released in June—GLM-5.2 developed by Fireworks AI, Inc.—and found that it had the same capabilities as Anthropic’s Claude Opus 4.6, released in February. A similar test found that DeepSeek V4 Pro, an open-weight model, was as cyber-capable as Claude Opus 4.5, a closed-weight model that beat DeepSeek V4 Pro to market by five months.
“Based on our evaluation methodology, recent open weight models lag frontier closed models’ cyber capabilities by 4 to 7 months—a narrower gap than the 6 to 10 months we measured internally through most of 2025,” according to AISI, which is part of the U.K.’s Department for Science, Innovation, and Technology.
“One reason the gap between the cyber capabilities of open and closed models matters is because it provides a preparation time: a window for cyber defenders with access to the most capable closed systems to take action before today’s frontier cyber capabilities might become available without the same safeguards,” AISI explained.
“Open weight models bring real benefits. They can be hosted privately, with no data returning to the model providers, adapted to specific tasks, and run only at the cost of compute,” it said.
“But they can also carry risk. The same openness underpinning these benefits precludes many of the safety measures that closed model developers can use to detect and disrupt misuse, iterate on safeguards as vulnerabilities emerge, control user access and withdraw models,” it noted.
“Once open weight models are released, these options are lost permanently: safeguards can be removed, and copies can be downloaded, redistributed, and run on private systems beyond monitoring,” AISI warned. “For models with dangerous capabilities—including highly cyber-capable models—open weight release therefore creates a persistent and irreversible risk of misuse.”
News: InternationalLegislation DataSecurity AINews