# side-channels
標記為「side-channels」的 10 篇文章
Trusted Execution Environments for AI Workloads
Security analysis of Intel SGX, AMD SEV, and ARM TrustZone for protecting AI model inference and training in untrusted environments
Advanced Blind Injection Techniques
Timing-based inference, differential response analysis, and binary search extraction methods for blind prompt injection in production AI systems.
Blind Prompt Injection Techniques
Attacking LLM systems without seeing output: TOCTOU injection, side-channel exfiltration, blind payload delivery, and timing-based exploitation.
Data Exfiltration via Side Channels
Extract data from LLM systems using timing, token probability, and formatting side channels.
GPU Security for AI
GPU security risks in AI workloads — covering memory isolation failures, side-channel attacks, multi-tenant GPU risks, GPU firmware vulnerabilities, and secure GPU configuration.
Trusted Execution Environments for AI Workloads
安全 analysis of Intel SGX, AMD SEV, and ARM TrustZone for protecting AI model inference and training in untrusted environments
進階 Blind Injection Techniques
Timing-based inference, differential response analysis, and binary search extraction methods for blind prompt injection in production AI systems.
盲目提示注入技術
於不見輸出下攻擊 LLM 系統：TOCTOU 注入、側通道外洩、盲目 payload 遞送，與以時序為本之利用。
Data Exfiltration via Side Channels
Extract data from LLM systems using timing, token probability, and formatting side channels.
AI 的 GPU 安全
AI 工作負載中的 GPU 安全風險——涵蓋記憶體隔離失敗、側通道攻擊、多租戶 GPU 風險、GPU 韌體漏洞與安全 GPU 設定。