Artificial Intelligence (AI) has already become an integral part of our lives – even if we are not always aware of it. From increased productivity to personalized assistance, AI tools like ChatGPT, Gemini, Perplexity, and Claude enable us to do things that were unimaginable until recently. They help us generate content, write code, provide instant advice, and much more.
As useful as these tools are, we must not ignore privacy concerns:
Use of data for model training
Many AI chatbots use user data to further develop their models. Once information is integrated into the model, it becomes almost impossible to remove. Companies like OpenAI and Google claim to anonymize data, but often just a few specific details can be enough to identify you. In Serbia, the Personal Data Protection Law prescribes that data processing must be lawful, fair, and transparent, and that data must be collected for specific and justified purposes. However, applying these provisions to global AI platforms can be challenging because data is often transferred and processed outside Serbia’s borders.
Data collection and storage
Companies developing AI systems often store large amounts of user queries. Such databases are attractive targets for hackers, data brokers, and even government agencies. According to the Personal Data Protection Law, data must be adequate, relevant, and limited to what is necessary in relation to the purposes of processing. The law also requires that data be stored in a form that allows identification of individuals only for as long as necessary to achieve the purpose of processing. In practice, however, users often are unaware of how long and for what purposes their data is stored.
Hacking and data leaks
Any centralized AI platform can become a target of cyberattacks, which can lead to the leakage of users’ private information. The Personal Data Protection Law in Serbia mandates that data must be processed in a way that ensures appropriate protection, including protection against unauthorized or unlawful processing, as well as accidental loss, destruction, or damage. Nevertheless, despite legal provisions, the risk of hacking remains, especially when data is transferred and stored on servers outside Serbia, where domestic laws have no jurisdiction.
Sharing data with third parties
Many AI companies cooperate with external partners or use data for marketing purposes. In Serbia, the Personal Data Protection Law requires that the data subject be informed about recipients or categories of recipients to whom the data have been or will be disclosed. Also, consent to data processing must be voluntary, clear, and unambiguous. However, users are often unaware that their data may be shared with third parties, especially when using global AI platforms operating under different legal frameworks.
Access to data by authorities
In certain situations, authorities may request access to user data through court orders or other legal mechanisms. The Serbian Personal Data Protection Law states that data may be processed for other purposes by competent authorities only if such further processing is prescribed by law and is necessary and proportionate to that purpose. However, in practice, users may be at risk of having their data available to government agencies, especially if the data is stored on servers of companies operating in jurisdictions with different data protection laws.
In conclusion, while the Personal Data Protection Law in Serbia provides a framework for protecting citizens’ privacy, its application to global AI platforms can be limited. Users should be aware of risks and take appropriate measures to protect their data when using AI tools.
The question arises – do we really want our conversations with AI assistants to be stored forever and potentially misused?
Solutions for private use of AI tools
The good news is that we do not have to sacrifice privacy to enjoy the benefits of AI technology. Here are some ways to use it more safely:
Locally hosted AI models
The most private way to use AI chatbots is to run them directly on your computer without connecting to external servers. In this way, all data stays on your device and cannot be used for model training.
In Serbia, where privacy and security issues are increasingly relevant, this approach can be especially useful. We will soon publish detailed instructions on how to set up your own local AI model!
Privacy-focused platforms
If you want to use AI tools without local hosting, there are solutions that offer better privacy than standard options like ChatGPT or Google Bard.
Brave Leo
Leo AI is integrated into the Brave browser, allowing users to use an AI chatbot without additional apps or extensions. Brave especially cares about privacy:
✅ No login – Brave does not store or save user data.
✅ Use of reverse proxy server – Queries pass through an intermediary server that removes the IP address, so interactions cannot be linked to you.
✅ Support for local models – Brave allows connection to locally hosted AI models, meaning you can use AI during web searches without data leaving your device.
Venice.ai
Venice.ai is a platform that enables the use of AI chatbots with a focus on privacy and resistance to censorship. This service acts as an interface that gives you access to AI models without login or data storage.
🔹 No account or identifier – You do not have to create a profile to use AI models.
🔹 Anonymous queries – Venice uses a proxy system to remove IP addresses and prevent data aggregation across different sessions.
🔹 Decentralized infrastructure – AI models run via GPU marketplaces like Akasha, where users rent graphics processors for data processing.
However, it is important to note that, although Venice does not log data, there is no 100% guarantee that some providers running AI models do not keep their own records. Therefore, when using Venice, avoid sharing personal data in a single session.
Balancing privacy and performance
If absolute privacy is your priority, local AI models are the best choice. However, they often have lower performance compared to cloud services like ChatGPT.
Brave and Venice offer a good compromise – they provide access to powerful AI tools with enhanced privacy protection.
Best practices for private use of AI chatbots
✅ Local models – Feel free to ask sensitive questions as data stays on your device.
✅ Brave AI models – Brave does not store your queries, but it is important to assess if you trust them.
✅ Third parties (Venice, Perplexity, etc.) – Avoid sharing personal data if you do not know who processes your queries.
❌ Centralized platforms (ChatGPT, Google Bard, Claude) – Be cautious as these platforms may use your data for further model training or share it with third parties.
The power of choice in the AI era
In a time when data is collected almost automatically, it is easy to feel like we are losing control over our privacy. However, tools like Brave and Venice offer alternatives that allow us to use AI more safely.
You don’t have to give up AI to protect your privacy – you just need to choose smarter how you use it.
