Ai Tool Treatment
UCI Information Security broadens institutional AI security guidance to cover agentic AI tools alongside chatbots, mandating caution over system data access, and lists non-UCI Anthropic Claude as approved for Protection Level 1 (P1) data only.
标准化值: {"scope": ["chatbots", "agentic_tools"], "agentic_guidance": "treat_system_data_access_with_equal_risk_caution", "allowed_tools_data_levels": {"UCI ZotGPT": ["P1", "P2", "P3"], "UCI Microsoft Copilot": ["P1", "P2", "P3"], "UCI Google Gemini": ["P1", "P2", "P3"], "OpenAI ChatGPT": ["P1"], "Microsoft Copilot (Non-UCI)": ["P1"], "Google Gemini (Non-UCI)": ["P1"], "Anthropic Claude (Non-UCI)": ["P1"]}, "p4_rule": "default_unacceptable_without_exception_review"}
原始证据
证据 1Similarly, be aware of what data any Agentic AI tools you are using can access on your systems, and treat that with the same risk level and caution as if it was being entered into a chatbot... Below is a list of services and the appropriate data that can be used with them based on protection level... AI Service Allowed Data Protection Level(s)... Anthropic Claude (Non-UCI) P1
本地化显示 only
Similarly, be aware of what data any Agentic AI tools you are using can access on your systems, and treat that with the same risk level and caution as if it was being entered into a chatbot... Below is a list of services and the appropriate data that can be used with them based on protection level... AI Service Allowed Data Protection Level(s)... Anthropic Claude (Non-UCI) P1



