The product doesn't even require a VPN to work. Go ahead and test!
Brave introduced its proprietary AI assistant "Leo", which is able to perform most of the tasks that chatbots usually handle, but with an increased emphasis on maintaining privacy.
The novelty is already available in the Android version of the Brave browser in the latest update 1.63, as well as in desktop versions of the browser, including Windows, Linux and macOS. They promise to add iOS support a little later.
Leo is capable of performing a wide range of tasks, including short retelling of information on web pages and videos, answering questions about submitted content, translating pages, writing code, transcribing video or audio clips, and generating text content.
To call the assistant on a mobile device, just open the browser menu by pressing the "⋮ "button and select "Leo" there. In desktop versions, just type your request in the browser's address bar and select "Ask Leo a question".
Brave claims that Leo stands out from other AI assistants for its ability to provide high-quality and relevant answers with fewer errors thanks to the use of several advanced language models, such as Mixtral 8x7B, Claude Instant and Llama 2 13B, which is used by default by Mixtral.
The assistant works in Russia without a VPN and understands Russian, although for some reason he answers mainly in English, even if you ask him to do it in Russian. This may depend directly on the selected LLM model.
As mentioned above, by default, the assistant responds using the advanced Mixtral AI model, but it works slowly in the free version, and the request limit is triggered extremely quickly.
After the limit expires, the assistant offers the user to either subscribe for $15 / 1,390 rubles per month, or switch to a simpler free model, which, by the way, works much faster when compared with the Mixtral trial version.
On Android, a subscription is issued through Google Play, so you can't connect it without dancing with a tambourine even if you really want to. You may be able to circumvent this restriction by registering it from your computer.
The advent of AI tools on mobile devices raises concerns about data privacy, as these tools often run in the background, monitoring all user interactions with content and services.
Brave says that "Leo" is fundamentally different in this regard, including strict security measures that guarantee privacy, while allowing you to use the full power of AI tools.
Among the security measures reported by developers, we can distinguish the following:
As already noted at the beginning, the mobile version of "Leo" is available so far only on Android, and even then, perhaps, not for everyone, since the deployment is gradual. Our editors were lucky, so we managed to test the chatbot right away. The iOS version is expected to be released within a few weeks. As noted earlier, desktop versions should work for all users at once.
The AI assistant "Leo" demonstrates that artificial intelligence technologies and privacy protection do not contradict each other. Moreover, their combination opens up new opportunities for users, making products both more useful and safer. This is an important step towards the responsible use of AI in the public interest.
Such innovations make a valuable contribution to shaping a more transparent and ethical future in the use of artificial intelligence.
Brave introduced its proprietary AI assistant "Leo", which is able to perform most of the tasks that chatbots usually handle, but with an increased emphasis on maintaining privacy.
The novelty is already available in the Android version of the Brave browser in the latest update 1.63, as well as in desktop versions of the browser, including Windows, Linux and macOS. They promise to add iOS support a little later.
Leo is capable of performing a wide range of tasks, including short retelling of information on web pages and videos, answering questions about submitted content, translating pages, writing code, transcribing video or audio clips, and generating text content.
To call the assistant on a mobile device, just open the browser menu by pressing the "⋮ "button and select "Leo" there. In desktop versions, just type your request in the browser's address bar and select "Ask Leo a question".
Brave claims that Leo stands out from other AI assistants for its ability to provide high-quality and relevant answers with fewer errors thanks to the use of several advanced language models, such as Mixtral 8x7B, Claude Instant and Llama 2 13B, which is used by default by Mixtral.
The assistant works in Russia without a VPN and understands Russian, although for some reason he answers mainly in English, even if you ask him to do it in Russian. This may depend directly on the selected LLM model.
As mentioned above, by default, the assistant responds using the advanced Mixtral AI model, but it works slowly in the free version, and the request limit is triggered extremely quickly.
After the limit expires, the assistant offers the user to either subscribe for $15 / 1,390 rubles per month, or switch to a simpler free model, which, by the way, works much faster when compared with the Mixtral trial version.
On Android, a subscription is issued through Google Play, so you can't connect it without dancing with a tambourine even if you really want to. You may be able to circumvent this restriction by registering it from your computer.
The advent of AI tools on mobile devices raises concerns about data privacy, as these tools often run in the background, monitoring all user interactions with content and services.
Brave says that "Leo" is fundamentally different in this regard, including strict security measures that guarantee privacy, while allowing you to use the full power of AI tools.
Among the security measures reported by developers, we can distinguish the following:
- Users do not need to log in to a Brave account to use "Leo" with the free LLM model. There is no risk of profiling or deanonymization;
- "Leo" does not store chats with users and does not use them for training / optimizing models;
- All user requests are transmitted through the anonymization server, so user IDs and their requests are not linked;
- "Leo" responses are deleted from Brave servers after they are created. Responses that use models from Anthropic are deleted after 30 days;
- No user IDs, such as IP addresses, are collected or stored;
- Leo Premium users who subscribe receive anonymous tokens confirming the status of their account, which are not linked to their payment details.
As already noted at the beginning, the mobile version of "Leo" is available so far only on Android, and even then, perhaps, not for everyone, since the deployment is gradual. Our editors were lucky, so we managed to test the chatbot right away. The iOS version is expected to be released within a few weeks. As noted earlier, desktop versions should work for all users at once.
The AI assistant "Leo" demonstrates that artificial intelligence technologies and privacy protection do not contradict each other. Moreover, their combination opens up new opportunities for users, making products both more useful and safer. This is an important step towards the responsible use of AI in the public interest.
Such innovations make a valuable contribution to shaping a more transparent and ethical future in the use of artificial intelligence.
