We’re launching Lyria 3.5 in Google Flow Music, with advances across musicality, lyrics, vocals, and creative control
deepmind.google·6d ago
AI Engineering
Hacker News·Jun 28·5 min read
A way to exclude sensitive files issue still open for OpenAI Codex
TL;DR
Sensitive data and large irrelevant files can inadvertently be sent to AI models, posing privacy risks. A Rust implementation, codex-rs, was proposed to address this issue, but as of August 2025, it lacks the necessary features.
✦ Why It Matters
Engineers can prioritize data privacy and efficiency by implementing file exclusion features in AI models.
Key Takeaways
- Explore Rust implementations for AI data management.
- Engage in discussions about file exclusion features.
- Assess the impact of irrelevant data on model performance.
- Develop user-friendly interfaces for file selection.
Full Summary
AI models, like OpenAI Codex, can unintentionally process sensitive information or large files that are not relevant to the task at hand. To mitigate this risk, a Rust-based implementation called codex-rs was suggested, which would allow users to exclude specific files from being sent to the model.
However, as of August 2025, codex-rs does not yet support this functionality, leaving a gap in data privacy and efficiency. The goal is to restart discussions around this feature to develop a robust design that can effectively filter out sensitive and irrelevant files.
This could involve creating a user-friendly interface for file selection and exclusion criteria. The implications of successfully implementing this feature would enhance user trust and improve the overall performance of AI models by focusing on relevant data.
Engineers and researchers can benefit from understanding the importance of data management in AI applications.