Skip to content
@AI45Lab

AI45 Lab

Welcome 👋

to AI45, a safety ecosystem platform developed by Shanghai Artificial Intelligence Laboratory.

Core Philosophy

The platform is guided by the AI-45° Law. From a long-term perspective, AI safety and performance should ideally advance in parallel along a 45° line. Short-term fluctuations are permissible, but in the long run, this balance should neither fall below 45° (as at present) nor exceed it (to avoid constraining development).

Multiple technical pathways may achieve this “AI-45° Law”. We are exploring a causality-centered approach—“the Causal Ladder of Trustworthy AGI"—spanning three progressive layers: Approximate Alignment Layer, Intervenable Layer, and Reflectable Layer.'

Core Modules

🔬 Safety Foundation

🛡️ Safety Technology

🏆 Safety Evaluation

🌐 Safety Services

Popular repositories Loading

  1. ActorAttack ActorAttack Public

    Python 104 8

  2. Awesome-Trustworthy-Embodied-AI Awesome-Trustworthy-Embodied-AI Public

    JavaScript 65

  3. REEF REEF Public

    The repository of the paper "REEF: Representation Encoding Fingerprints for Large Language Models," aims to protect the IP of open-source LLMs.

    Python 63 7

  4. Flames Flames Public

    Flames is a highly adversarial benchmark in Chinese for LLM's harmlessness evaluation developed by Shanghai AI Lab and Fudan NLP Group.

    60

  5. CodeAttack CodeAttack Public

    [ACL 2024] CodeAttack: Revealing Safety Generalization Challenges of Large Language Models via Code Completion

    Python 53 8

  6. VLSBench VLSBench Public

    [ACL 2025] Data and Code for Paper VLSBench: Unveiling Visual Leakage in Multimodal Safety

    Python 51 1

Repositories

Showing 10 of 32 repositories

People

This organization has no public members. You must be a member to see who’s a part of this organization.

Top languages

Loading…

Most used topics

Loading…