Hi everyone,
I’m looking for recommendations on free (or freemium) AI assistants that meet these criteria:
✅ Strong capabilities – Can handle complex questions and research. ✅ Privacy-respecting – Doesn’t train on user data, preferably E2E encrypted. ✅ Ethical stance – Not complicit in oppression or human rights violations (Palestine, sudan, uygur, etc). ✅ Not on BDS list – Companies that aren’t supporting or profiting from occupation.
Which AI tools have you found that respect both privacy AND human rights?
I’d really appreciate honest recommendations and experiences. Trying to find tools that align with my values without sacrificing functionality.
Thanks in advance! 🤝
https://euria.infomaniak.com/shared/01a1cb91-3c62-7024-ab77-5e3c64e5ab63
Infomaniak is an independent Swiss technology company founded in 1994 in Geneva, majority-owned by its employees. It designs, develops, and operates its own data centers in Switzerland, hosting all customer data exclusively in Switzerland.
What it offers:
- Web hosting and online stores (with CMS support and automated backups)
- Domain name registration and management
- Cloud solutions (servers, VPS, Kubernetes, Jelastic)
- kSuite — a collaborative suite including email, cloud storage (kDrive), calendar, video conferencing (kMeet), team messaging (kChat), and an AI assistant
- Audio/video streaming solutions for media, radio, and events
Key commitments:
- Privacy: No advertising, no data mining, GDPR and Swiss law compliant
- Ecology: Data centers recycle 100% of consumed energy to heat local housing; 200% CO₂ emission compensation
- Quality & Security: Certified ISO 27001, 9001, 14001, and 50001; B Corp™ certified since 2025
- Sovereignty: 100% Swiss-made infrastructure, operated from Geneva and Zurich
There’s a tech incubator, i think its called “tech for palestine”.
They recommend thaura.ai, it was started by 2 syrians who felt there was a gap in this space.
I think there might also be qaf.ai which is more islamic oriented but definitely pro-pali.
Then there is also confer.to which was started by signal co founder that offers encrypted AI (meaning it uses fancy math that can work on encrypted data without decrypting it), though im not too familiar on their stance on palestine and other issues.
https://confer.to/blog/2026/01/private-inference/
They decrypt the data inside the VM, and run it in plaintext there. No fancy math. As far as I am aware, the dream of doing operations on encrypted data is still mostly theoretical. Still, worth considering, they are going a long way to ensure that the VM is un-tampered with, and Moxie Marlinspike generally knows their shit.
Self host is your only option. Companies can make all the privacy claims in the world, but nothing prevents them stealing your data. e2e encryption is meaningless when the other end is the company, and has to be able to read your inputs in order to inference with them. Anyone claiming e2e encryption is straight up lying to you.
If you are actually based in Palestine, it doesnt seem likely you would be able to purchase GPUs (and keep the running/safe), so you are more realistically looking at GPU rental, which can be pricy. And it may not be secure either.
A nvidia P100 can be had from aliexpress for relatively cheap, and gives you 16GB vram. You won’t be running fable level models, but there are models that will work. If your budget can stretch to 2, there is lots more options available. I dont know what kind of questions you want to ask, but I’m happy to throw a sample at my p100 system if you want?
I agree with what you said, but the hosting option isn’t available right now, and it’s not available to everyone either. So what I’m looking for is the least harmful option. Because, to my knowledge, there are models that don’t use user data to improve their services—they’re trained on either historical data or open-source data, or they rely more on web searches for their answers. (Many people recommend thaura.ai)
The smaller companies may be the best bet for not training on your data, because they cannot train models themselves usually. But there is basically no way to guarantee that they wont sell your data onwards to the bigger labs. Thaura.ai has a commendable privacy policy, but that is only worth as much as your ability to audit and sue them if they fail to meet it. So effectively, its worthless. You basically have to trust them, and given so many of these AI companies start and die (or get purchased) all the time, that’s near impossible to do.
If privacy is important, don’t send your data to any company.
