Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
willvarfar
31 days ago
|
parent
|
context
|
favorite
| on:
What GPT-OSS leaks about OpenAI's training data
You can imagine LLM fingerprinting to be part of future pentest workflows where they identify the model and know it's weaknesses and vulnerabilities etc...
varispeed
31 days ago
[–]
Will be cat and mouse game, because models can be trained to mimic response of other models under such conditions to obfuscate.
dboreham
31 days ago
|
parent
[–]
See also: browser user-agent field.
Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
