AISI
A government-supported AI safety research institution in the United States dedicated to conducting independent safety evaluations of frontier AI models and advancing the standardization of AI governance frameworks and practices.
Core Facts
The UK AI Safety Institute is one of the official bodies that relatively early conducted systematic evaluations of AI model security risks on a global scale.
AISI was established in November 2023 and is the world's first institution set up by a national government dedicated to frontier AI safety research and evaluation.
Timeline (last 90 days)
AISI has developed open-source evaluation tools such as Inspector.
An incident occurred in a recent AISI network test, involving some kind of unexpected behavior exhibited by the model during testing, with specific technical details not disclosed.
Existing third-party evaluation attempts include the evaluation agreements signed between the UK AI Safety Institute (AISI) and multiple labs, the US NIST's AI risk assessment framework, and the third-party compliance assessments stipulated in the EU AI Act.
The US government's AI Safety Institute (AISI) and major AI companies have incorporated red-team testing into the standard process prior to model releases.
All Facts (8)
The UK AI Safety Institute is one of the official bodies that relatively early conducted systematic evaluations of AI model security risks on a global scale.
90%VerifiedAISI was established in November 2023 and is the world's first institution set up by a national government dedicated to frontier AI safety research and evaluation.
80%UnverifiedAISI的评估采用多层次测试框架，参照CVE数据库中的真实案例，同时设计全新挑战场景测试模型泛化能力
85%UnverifiedAISI was initially under the UK Department for Science, Innovation and Technology (DSIT) and was later reorganized into an independently operating public body in early 2025.
60%UnverifiedAISI has developed open-source evaluation tools such as Inspector.
50%UnverifiedExisting third-party evaluation attempts include the evaluation agreements signed between the UK AI Safety Institute (AISI) and multiple labs, the US NIST's AI risk assessment framework, and the third-party compliance assessments stipulated in the EU AI Act.
50%UnverifiedThe US government's AI Safety Institute (AISI) and major AI companies have incorporated red-team testing into the standard process prior to model releases.
50%UnverifiedAn incident occurred in a recent AISI network test, involving some kind of unexpected behavior exhibited by the model during testing, with specific technical details not disclosed.
50%