These three alternatives want to change the rules

Artificial intelligence chatbots have crept into our daily lives without asking permission. We talk to them to work, resolve doubts or simply vent, but many times we overlook an important detail: services as popular as ChatGPT, Gemini or Copilot They use these conversations by default to train their models. And there they end up ranging from personal issues to work issues that perhaps we would never share with a stranger.

For anyone who values ​​privacy, this may be a cause for concern. Turning off training from the settings menu helps, but does not eliminate all risks. There is some possibility that a security breach could also expose conversations. So the question that we haven’t quite resolved returns again: are there alternatives that really put privacy first? And if there are, which ones work best?

AI chatbots that better protect our privacy

In a new video from Xataka’s YouTube channelFrancisco Franconi tries to shed light on this dilemma. He himself admits that he is not the best example of prudence: over the years he has turned sensitive information in ChatGPT. “He has my tax return, my medical history, he knows what time I walk my dog, how many days I have left on vacation,” he confesses.

But this time he is determined to change the rules of the game. After testing several options, it has selected three alternatives that focus on privacy and that also have free versions. He acknowledges that there are more, but he preferred to focus on those that really convinced him.

The first is Venice. “Your data is encrypted and are stored locally in your browser and not on its servers,” explains Franco, highlighting one of its big claims. The way Venice works is peculiar: it adopts a different mechanic than usual to manage messages, a process that he details step by step in the video and that helps to understand why it fits so well into a more secure approach.

The second proposal is the work of an old privacy acquaintance. “If you are a freak privacy, you most likely know the DuckDuckGo search engine, an alternative to Google that became famous for not track your searches or your history,” says our colleague before presenting Duck.AI. The tool, he assures, has been very well received and its philosophy is clear: “It does not compete by offering its own model, but rather guarantees privacy in the use of commercial or third-party models,” he adds.

In the video, Franco also shares practical recommendations for getting the most out of Duck.AI, which is useful if you are looking for a chatbot that fits into more intensive workflows.

The third option is, perhaps, the least known, but also the one that makes the most forceful declaration of intentions: Okara. For Franco, this platform is “a declaration of war on the data business model.” Its proposal is supported by its own infrastructure, open source models and an encryption approach that gives full control to the user. “Although your history is stored on their servers, it is encrypted in such a way that only you can decrypt it using your key,” he explains.

In addition to explaining how Okara works and what advantages it brings, Franco also talks about its limitations. For some users they will be minor details, but for others they can tip the balance towards one platform or another.

Each of these alternatives offers a different approach, but a common goal: privacy. And you, which one would you choose? Have you tried any or are you tempted to take the leap? We read you in the comments. And remember that you can see the full video in the Xataka YouTube channel.

Images | Xataka

In Xataka | DeepSeek has launched its new reasoner model. It’s free and beats GPT-5