The bigger PII risk in LLM apps isn't the provider — it's your logs, traces, and agent memory
abhi1510 · reddit · 2026-10-28
A Reddit discussion thread argues that when sending sensitive data to LLMs, too much focus goes to whether OpenAI or Anthropic can be trusted, while the bigger concern is the rest of the stack: prompts containing names, SSNs, medical or financial data can end up in application logs, observability tools, error traces, conversation histories, agent memory, eval datasets, and downstream API calls. The analogy: developers wouldn't put an API key in a prompt even if they trusted the provider — but you can rotate a key, not someone's SSN. He asks production devs whether they actively mitigate this via redaction, tokenization, anonymization, self-hosting, or contracts, and how it ranks on their engineering priority list.
More from Safety
- AI cyber-wreckage noise far outstrips actual damage so far, argues new analysis — timigod · 2026-10-28
- OpenAI to co-host webinar on php2shell disclosure and AI-driven vulnerability discovery — gdb · 2026-10-28
- Rust and Verus shrink the search space for AI-driven CVE hunting, but does it ever end? — jd_pressman · 2026-10-28
- Anthropic pushes back on Slack breach rumors but admits security must be reinvented yearly — logangraham · 2026-10-28
- jackson-databind hit with CVSS 9.8 flaw CVE-2026-110879, bypassing prior fix — Kyrannio · 2026-10-28
- Archil founder: AI reverse-engineering is making it unsafe to ship binaries to customers — RexDouglass · 2026-10-28