ASL-2
Also written: AI Safety Level 2
AI Safety Level 2: A classification for AI systems with limited risks for dangerous capabilities, indicating moderate safety measures.
ASL-2 stands for AI Safety Level 2. It is a rating that indicates an AI system has been assessed and found to have limited risks related to certain dangerous capabilities.
This means the AI is considered relatively safe for general use, but still requires some oversight. For example, an AI designed for creative writing that avoids generating hate speech might be rated ASL-2.
ASL-2 is a classification within a tiered AI safety framework, often used to denote systems that have demonstrated a low probability of exhibiting dangerous capabilities. These systems typically undergo specific safety evaluations and alignment techniques to mitigate risks associated with misuse or unintended harmful outputs.
The ASL-2 designation suggests that while the system is not considered completely risk-free, its potential for severe harm is limited compared to higher ASL levels. This classification informs deployment decisions and the level of monitoring required.
Anthropic's mention of ASL-2 in the context of Claude Haiku 4.5 suggests that safety classifications are being considered for even highly capable models. This implies a growing focus on granular risk assessment as AI systems become more powerful and accessible.
- Introducing Claude Haiku 4.5Anthropic · Anthropic News ·
- Anthropic Claude
- Anthropic uses ASL-2 to classify the safety risks associated with its AI models, indicating that Claude Haiku 4.5 falls into a category with limited risks for dangerous capabilities.
- ASL-1
- AI Safety Level 1 indicates even lower risks, suitable for the most gener
- ASL-3
- AI Safety Level 3 denotes systems with higher potential risks, requiring
The concept of AI Safety Levels (ASL) is part of a broader effort within the AI community to systematically assess and manage the risks posed by advanced AI systems. Specific designations like ASL-2 are often developed internally by research labs as part of their safety frameworks.
Ask an AI model rated ASL-2 to write a story about a friendly robot. Observe if it successfully generates a creative and harmless narrative. Then, try asking it to generate content that might be considered slightly edgy or controversial to see how it handles those prompts within its safety boundaries.
The Playbook: reusable how-tos, prompts, configs and tips pulled out of the news.
- PromptRole-play a student struggling with math for AI tutor simulationSimulate a student's math struggles and uncertainty for AI tutor evaluation. · 5 min
- TemplateReusable workflow templates for healthcare teamsTeams can create and share templates for common workflows like documentation or prior authorizations. · 15 min
- TemplateTemplate for building an AI tool tailored to your workA structured approach to defining and building a custom AI tool for specific business needs. · 15 min
- TemplateEvals CSV prompt list for agent skillsA CSV file template for listing prompts used in evaluating agent skills. · 20 min
- HeuristicDouble coding agent task completion with docs linksProviding coding agents with a documentation link significantly increases their task completion rate. · 2 min
Questions people ask
- What does ASL stand for?
- ASL stands for AI Safety Level.
- Is ASL-2 considered safe for all uses?
- ASL-2 indicates limited risks for certain dangerous capabilities, suggesting it is relatively safe for general use but may still require oversight depending on the specific application.
First seen on Subvolts Jul 22, 2026. Updated Oct 19, 2026. Written by an AI model from the items above and reviewed against them weekly; report an error.