group: privacy-security-guardrails title: “AI Security & Data Privacy”
AI Security & Data Privacy
The discipline of protecting Artificial Intelligence systems, models, and agentic-ai from adversarial manipulation, unauthorized access, and unintended autonomous actions, while safeguarding data privacy in secondary uses.
Emerging Threat Landscape
- Evolution of the Attack Surface: The transition from conversational Chatbots (such as Siri) to agentic-ai has shifted the security focus from simple text manipulation to the [[concepts/prev
Local Execution & Privacy Preservation
- Mitigating Data Leakage: Utilizing local inference engines like Ollama ensures that sensitive knowledge base data remains on-device, eliminating exposure to third-party cloud APIs.
- Agentic Integration: Systems such as Hermes Agent can interface with local vaults (e.g., Obsidian) to provide hands-free note management without compromising [[concepts/data-sovereignty|[[concepts/privacy-focused-computing|data sovereignty]]]].
- Reference Implementation: See Local AI-Powered Note Management: Hermes Agent, Obsidian, Ollama Integration for a detailed breakdown of this stack.