语音大模型定制时可能泄露敏感信息，因发音相似词被误识别。
When Helpful Context Leaks: Privacy Risks in Domain-Adapted ASR
- 通过提示词和微调方式定制模型，会引入隐私泄露风险。
- 在测试中，组合使用两种方式时泄露率显著升高。
- 避免使用上下文提示可降低泄露，兼顾准确率与安全。
语音大语言模型（SpeechLLMs）在专业场景中广泛应用，通常需通过用户提供的敏感上下文或专有录音进行领域定制。我们发现并系统研究了一种被忽视的隐私风险：当模型适应特定术语后，可能因发音相似而错误转录出上下文或训练数据中的词语，即使实际输入不同。为此，我们构建了受控数据集，评估了提示词与微调两种定制方式下的泄露率，结果显示两者均导致可测量的泄露，且联合使用时风险叠加。我们还测试了提示层缓解策略，并分析了不同定制方法下的准确率-泄露权衡，发现仅使用微调而不加上下文提示能取得最佳平衡。相关代码与数据集已公开。
原文摘要 · Abstract (English)
SpeechLLMs are increasingly deployed in professional settings where domain customisation is standard practice: users supply context in prompts with sensitive information, fine-tune on proprietary recordings, or both. We identify and systematically investigate an overlooked privacy risk of such customisation: a model adapted to recognise domain-specific terminology can be nudged into transcribing a phonetically similar word from its context or training data, even when a different word is spoken, thereby leaking private information. To evaluate this risk, we construct a controlled dataset and measure leakage rates across two customisation mechanisms, prompting and fine-tuning. Both mechanisms cause measurable leakage, compounding when combined. We evaluate a prompt-level mitigation strategy and analyse the accuracy-leakage trade-off across customisation approaches, finding that fine-tuning without context prompts offers the best balance. We release our code and dataset publicly.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。