The Threat Prompt Newsletter

The Threat Prompt Newsletter

Share this post

The Threat Prompt Newsletter
The Threat Prompt Newsletter
DeepSeek app, safe to use?
Copy link
Facebook
Email
Notes
More

DeepSeek app, safe to use?

Craig Balding's avatar
Craig Balding
Jan 31, 2025
1

Share this post

The Threat Prompt Newsletter
The Threat Prompt Newsletter
DeepSeek app, safe to use?
Copy link
Facebook
Email
Notes
More
Share

From a WhatsApp chat with a good friend earlier this week…

Good morning my friend! Jumping on the topical bandwagon… DeepSeek app, safe to use?

Good morning Chief, yeah - just assume they will train on your data and usage activity (safe assumption for most services!)

Ta. As long as it’s not training itself on the contents of my phone. And TBH, I wish they learned more from usage; the amount of time I’ve spent telling Gemini off.

As for the app itself, here’s an “at a glance” (shallow but factual) assessment: https://reports.exodus-privacy.eu.org/en/reports/com.deepseek.chat/latest/

It’s telling me to kill any cyber security experts I know. 🤷 Don’t open any emails that smell of almonds!

But i probably would in a fit of Homer Simpson scent hypnosis…”Allllmmmoondds..mmm”


And this morning, I read that the Italian Data Privacy office instructed the delisting of the Deepseek app from the Apple and Google app stores.

The reason?

Insufficient information in response to their questions about how the service processes user data and what data was used to train the model (and what permissions were sought from data subjects).

I’m using hosted Deepseek models with my AI pair programmer (aider) for coding via OpenRouter. In that context, my first impressions are positive. Plus, OpenRouter gives me the option to route all my inference requests to a Deepseek provider that makes privacy promises I prefer.

P.S Did you clock the comment from my friend? “I wish they learned more from my usage”. I suspect heavy AI users would happily pay for this to be the case. I know I would since it would reduce roundtrips, which saves time and tokens ($$).


Subscribe to The Threat Prompt Newsletter

By Craig Balding
Get Daily AI Cybersecurity Tips
Nayan's avatar
1 Like
1

Share this post

The Threat Prompt Newsletter
The Threat Prompt Newsletter
DeepSeek app, safe to use?
Copy link
Facebook
Email
Notes
More
Share

Discussion about this post

User's avatar
TP#18 The AI Trust Paradox
Plus: High Impact Prompt Injection through ChatGPT Plugins
May 20, 2023 • 
Craig Balding
1

Share this post

The Threat Prompt Newsletter
The Threat Prompt Newsletter
TP#18 The AI Trust Paradox
Copy link
Facebook
Email
Notes
More
TP#22 Uncovering Model Weaknesses with Garak
Plus: Etched in Tokens: Exploring LLM Watermarks
Jun 18, 2023 • 
Craig Balding
1

Share this post

The Threat Prompt Newsletter
The Threat Prompt Newsletter
TP#22 Uncovering Model Weaknesses with Garak
Copy link
Facebook
Email
Notes
More
TP#21 Prompt Injection Snake Oil
Plus: Can you trust ChatGPT's package recommendations?
Jun 11, 2023 • 
Craig Balding
3

Share this post

The Threat Prompt Newsletter
The Threat Prompt Newsletter
TP#21 Prompt Injection Snake Oil
Copy link
Facebook
Email
Notes
More

Ready for more?

© 2025 Craig Balding
Privacy ∙ Terms ∙ Collection notice
Start writingGet the app
Substack is the home for great culture

Share

Copy link
Facebook
Email
Notes
More

Create your profile

User's avatar

Only paid subscribers can comment on this post

Already a paid subscriber? Sign in

Check your email

For your security, we need to re-authenticate you.

Click the link we sent to , or click here to sign in.