trained a prompt injection detector using ml-intern and DeepSeek v4 Flash, runs in the browser

“`html

• A British AI writer has trained a prompt injection detector using ml-intern and DeepSeek v4 Flash, which runs in the browser.
• The model, named “prompt-injection-detector,” achieved an F1 score of 95.87% on its initial training with DistilBERT targeting CPU inference.
• For future improvements, a larger synthetic dataset from Bordair was used and retrained the model to achieve better performance (F1 score of 99%).

“`

Stay ahead of AI. Get the most important stories delivered to your inbox — no spam, no noise.

Please enable JavaScript in your browser to complete this form.

Name *

First

Last

Name Email

Email *

Submit