Why are you shoving this AI garbage down our throats? If I wanted an AI assistant, I would install one. I do not want one. I do not want it to learn from my data. I don’t want it taking up space or using my CPU. I don’t want it making suggestions or hovering over my shoulder all the time. Please give us an option to remove it from the browser completely.
8 Likes
While I have the technical ability to do this, I’m guessing a lot of people don’t. This should be as simple as any other option in the settings, or better yet, make it a plugin you have to install if you want it.
6 Likes
This is something Brave has said they don’t want to do with any of their products. People have long been requesting it for Wallet, Rewards, VPN, etc. And now some are saying for Leo. They see it as vital to build it in as part of the browser. By having it as something like an extension, they said it creates multiple vulnerabilities for attacks by malware and all.
That said, it doesn’t gather our info or do anything unless we enable it. Even then, it’s all privacy preserving and doesn’t do much.
It doesn’t learn from your data. You may want to read through some of the things shared, such as links below:
The only feedback/training from users in regards to Leo is when people decide to provide feedback. Such as when an answer is given to you, there’s a hamburger menu on the reply you can click on, as showed in the screenshot below:
If you do something like Dislike Answer, you get a prompt like below:
If you then click Add Feedback, it shows like screenshot below:
That’s it. Otherwise it’s auto discarded and nothing saved or trained.
1 Like
AI without data is just a static program. I have been a software engineer for 40 years and have dabbled with AI, ML, etc. for decades.
1 Like
What is the policy to disable the AI? I don’t see anything that mentions Leo or AI in the brave://policy list. The web page you mention has BraveAIChatEnabled as a policy but setting that to 0 or 1 makes no difference.
1 Like
@ThatAintWorking,
Can you tell me how you configured the policy? It is admittedly not intuitive but I can assist you if I know how you went about creating the policy. I suspect it might be in the wrong directory, as I’ve done this myself. This would explain why it does not appear in the brave://policy list.
It should be an option one can decide to INSTALL and not a tech-class to get rid of it in a way which just hides this monster but it still works in the backgrounds and steals our search requests !
1 Like
The option should be that one who wants it can install it - and not that its there and one most remove it !
1 Like
Leo is already opt-in, it doesn’t communicate with Brave until you accept the privacy terms if I am not mistaken. Brave comes with quite a few features that not everyone wants, most of them embedded in the browser but opt-in to be enabled. It has a torrent client, IPFS, an adblocker, Tor, VPN, Leo, multiple wallets, their own ad network, and probably more that I’m forgetting. You kinda know what you’re getting into if you install Brave, that’s their business model: they show you these things and put them a click away, and hope to earn money out of your use of them either directly via subscriptions (e.g. VPN, Leo) or indirectly by enabling you to use services that they advertise on Brave (e.g. web3 related). There’s nothing special about Leo compared to all the other things. Switching to a model where all of these features require the user to go out of their way to install them is very easy to do technically but would mean Brave needs to find an alternative business model for making money. If you believe there’s a viable alternative then the source code is open, there’s nobody stopping you from opening a company making such an alternative browser.
AI SHOULD BE OPT IN NOT OPT OUT.
I do not want AI. It is not useful. It makes the internet worse. How do I turn it off.
PERMANENTLY not just on a single tab of search results. I NEVER want AI in my search results period.
This is coming from someone who has actively developed LLM based products.
Get rid of it or give me a way to permanently disable it.
How can I turn it off. Forever. And ever.
3 Likes
Just add
search.brave.com###llm-container
to your custom adblock filters in the settings (brave://settings/shields/filters) and you won’t see it on brave search anymore.
But it cannot be “forever”, sometime the CSS will change enough for the filter to stop working, just ask again then. Websites are not fixed versions of software, you know that when you start using any website: it will update whenever the owner feels like it.
1 Like
Eat it or die - huh ?`First theý started with a nice little browser - like most software - to get you used to it and then they force you into more and more THIS WORD IS NOT ALLOWED -CENSORSHIP TOO ?? – some dont notice it and even if so then you only have the choice - love it or leave it … is it this what you wanted to tell me ? I dont want an AI no matter what you assume … its there and I dont want it there … .I dont want all the options for datastealing pre-activated as they are when you open a private site or a TOR-site – they tell you to respect your privacy but steal from you and I think that most users think when they deactivated the options in the main settings that they are deactivated everywhere - but they are not ! they are fully activated in the settings of private sites and TOR-sites ! They steal userdata from users who dont know that their data are stolen because they deactivated them in the main settings !
Maybe Leo is “opt-in” when it comes to this horrible chatbots but its not an option when it comes to the search-engine ! There this disgusting creature is there and always pops up when you type in a search request – it collects our search requests and behavior and there is no option to get rid of it -.try it out before you tell people to eat or die !
“cannot see it” - or is it really totally deactivated and blocked and cannot follow my activities anymore ? I do not allow these creature to be trained by my behavior - and this is what they need - to learn from user behavior and this is what I want to avoid no matter what it costs ! So the best is not to include it in the first time - take it out and only offer it for download for those who want it but I bet these will not be many - thats why you force it on users !
added the code - reloaded - the AI icon is still “red” when I start typing a search-request in the search-field - its there - the monster is there !
@ajaccio
Thank you for the feedback.
To anyone else in this thread — for disabling Leo entirely such that it will not appear anywhere in-browser, we are working on easier more manageable policy options at the present time. More information on this soon.
In the mean time, you can remove Leo completely by doing the following:
- Go to
Settings --> Leoand disable both toggle options here:
- To remove it from autocomplete suggestions, go to
Settings --> Appearanceand ensure that theLeo AI Assistantoption is unchecked:
You will now see no more in-browser options/UI for Leo, with the exception of the actual Settings --> Leo option on the left-hand side of Settings.
For users who would like to remove Leo from the Brave Search site (note, this is separate from the browser itself):
- Visit search.brave.com
- Click the “gear” icon on the top-right to open settings, then
Show more - Scroll down and disable the
Answer with AItoggle
Lastly, if the tiny icon in the search bar on search.brave.com continues to bother you (note that this button does absolutely nothing unless directly interacted with):
- Visit search.brave.com
- Right-click anywhere on the page, select
Brave --> Block element - Hover the cursor over the AI button and click on it
- Click
Createin the box at the bottom right of the page
Thank you
3 Likes
The first thing I do when I install any new software - I go and deactivate EVERYTHING which is pre-activated – because there is always a reason why its preactivated - so no AI was ever activated by me … and if there was anything preactivated like “sending anonymous data” and so on I deactivated right after I installed Brave – but its still preactivated on Private Sites and TOR sites and its not possible to deactivate it before the first search !!! So I am sending all my data including my nation, my user-behavior etc against my will and main-settings and am feeding this disgusting AI with my search on private sites – so my “private” is not private at all – and I dont care if its “anonymous” or not - I will never allow that my behavior is feeding and training AIs or other datamonsters ! Either it is private or not – so the first thing is that you set all options on “deactivated” and only who wants can activate them and not vice versa and you offer monsters like AI for those who want it after a readable explanation how dangerous it is and dont force it on everyone with this scam-option for deactivating which is still collecting my behavior !
Even when you deactivate everything in your main settings, go to " all regions" and turn off “moderate” and most of all deactivate all preactivated settings – as soon as you open a Private Site or a Private Site with TOR all the preactivated settings are active … and you cannot deactivate them before you do your first search - then only the little icon for “quick settings” shows up where you can deactivate them again and reset your national position which always shows up there … so you send all your data to Brave AND the AI although you thought that you deactivated all of this in the main settings … and the worst crime is that its on Private Sites – where you think you not only deactivated everything but are extra private - so you send your most private behavior to this criminal AI to feed and train this monster ! Its time that we enforce a law that nothing ever must be pre-activated and opted in … we need a law where nothing is sent or one is not opted in into anything before one reads a clear explanation what its doing and what kind of risks are in there … yes its possible if we all work together ! Most of all to fight the AIs before they can take over !
2 Likes
“Die”? In what sense do they even remotely threaten to do something bad if you don’t “eat” it? Leo is OPT-IN in the browser, you don’t have to accept the terms and it won’t do anything.
How are they forcing you exactly? You don’t have to upgrade your browser, you don’t have to keep using this browser, you don’t have to even use a browser. Nobody is forcing you to do anything. Plus all features can be disabled AFAIK, can they not?
Wrong again, those are not the only two choices. You don’t have to love it to use it.
I barely understand what you are trying to say but no this isn’t what I was trying to tell you. I said you don’t have to use features you don’t like, but you cannot expect a company to change their business model to accommodate for your weird requirement of not even downloading code that you don’t need. ESPECIALLY when the company is making an open source product. If you don’t like it, take the code and change it, or hire someone to do so. When you get it for free you get what they give you, they pay to make the code, so they chose what goes into it, not you. You are way too entitled.
Which option for datastealing is pre-activated? Leo doesn’t send page data until you accept the terms, and even warns you about it quite visibly.
Got any evidence for that or are you just making stuff up at this point? Cause AFAIK allmost all features are hard-disabled in private tabs and especially in Tor tabs. Leo cannot even be used on Tor tabs for example.
It would be nice if they had the option to disable it in the search engine with a checkbox in the options, I’ll give you that but your complaint is kinda unrelated to everything else you’ve accused them off and it’s presented as if it proves that they try to steal your data without an option to opt out, which is a lie because the search engine ALREADY sees every search query and every result you get back. If they want to steal your data they can do it whether they serve AI results to you or not. And before you say “oh but I only agreed to give my data to Brave, not to Anthropic (assuming they use Anthropic’s AI to give you result summaries)” I will remind you that your brave search data is passing through Amazon Cloudfront if I am not mistaken, and any datacenter provider that Brave uses technically has access to your data too. So adding AI to the search doesn’t change much in terms of how your privacy is affected. Just use the adblock filter I posted and you won’t have to see it, but they’ll still run it in the background but I explained how this doesn’t make a difference to your privacy. If you are threatened by something like this, you should be masking your IP and using Tor Browser.
This is FUD at this point unless you provide some actual evidence. Either you trust their search engine or you do not. Your search query and your IP is already in their hands once you perform a search using their engine. There is NOTHING special that an LLM can do to invade your privacy any more than how much they (Brave, Amazin Cloudfront, datacenter providers) already can technically invade it.
Maybe don’t move the goalposts? You were complaining about the chatbot and once I told you it’s opt-in you switched to the search engine which is a whole different story since there’s technically NO privacy protection since you already SEND your query to them and your IP to get service. The ONLY protection in a search engine is legal protections based on their privacy policy. Either you trust that, or you don’t trust it and you make do with IP masking. There’s NO threat model you could have where providing LLM summaries on the search page is insecure but not providing LLM summaries is secure enough, unless for some reason you don’t mind Amazon being able to see your search queries (with your IP necessarily attached to them since they are the CDN in front of Brave Search) but you mind Anthropic seeing queries, probably without IPs attached to them since they are irrelevant to the summarization process.
See my previous explanation about your threat model being inconsistent and therefore useless. Amazon’s CDN poses a greater risk that the LLM in the search engine.
Good luck navigating life in the next few years with this goal. This is similar to someone saying I do not want cars to be used to transport stuff and then expecting that they can do so by walking instead of taking a taxi, all the while buying stuff at the supermarket that have been transported there by trucks. Your search queries are some of the most useless training data you can find to help develop LLMs. What will they do with them? It’s just a search query. You sound like you think that the LLMs are trained based on specific user behavior, like “This IP searches often for “survival gear” therefore when you see other queries from that IP you should assume they are a prepper and speak to them in a way to make them afraid of the end of the world and manipulate them to buy stuff for that”. That’s not how LLMs are trained right now, and IF Brave wanted to do something like that, they can already do it without you SEEING the LLM summary. Again, your threat model is completely whack.
This is irrational. And I don’t mean that it is irrational to fear LLMs, I mean that the criteria that you use to figure out if you are threatened by something are completely wrong. To give you an example of what I mean: If you were afraid of dogs biting you, that’s fine, we can discuss whether it’s an irrational fear or not, but in the end you do you, and you can be afraid of dogs biting your all you like, and avoid dogs. But if you start complaining that you don’t want to go into a shop owned by someone who has a dog at home, or named after his dog, then you are behaving irrationally.
On a side note, how do you know that I haven’t already asked Leo for a summary of your posts here? I can highlight them and right click and select to summarize them. And it’s a public forum so it’s certain that companies are already using your messages to actually train LLMs since your messages, in contrast to your search queries, are actually proper prose. So how do you propose to use the internet when you are so hesitant to help train LLMs? You cannot even talk to any customer support over text nor phone because those will be LLMed too (some already are).
1 Like
- Eat it or die - friß oder stirb - is a popular quote !
- If I tell you it is the way I described it then it is the way - I would not even have to “proof” it if it would be possible. Of course I have screenshots … now show me how to upload them here – you know everything better so you surely also know this
And then there would be the simple way that you go and check it our by yourself – go to the search-site, type in a search-request and even when the AI-icon is deactivated it suddenly pops up red while you type and an alert under the search field shows up in red with an arrow upwards telling you - use the AI !!
After you use “enter” both red AI-forces vanish again and the AI item shows “deactivated” again …
Which means – you feed this horrible monster with your search requests and only dont see the AI-snippet results but the monster is active all the time !
And then go and set all preactivared options in your basic settings on “deactivated” - then go to “Private site” and go to “settings” there – it leads you back to your basic settings where everything is deactivated … now people feel safe …
BUT
go and search something … and when the results show up you have this little icon in the right upper corner for speed settings … go there and look - EVERYTHING IS ACTIVATED - and this on a private site where people feel extra safe most of all when they check it after opening the site where it leads back to the basic settings … .this is this is doubly bad because it’s a double scam - not only are you supposed to be safe on the private pages, but you’re being ripped off there in terms of data, but you’re also being misled because the settings don’t refer to the private pages, but lead back to the base page - and of course the system doesn’t remember the settings on the private pages, you have to repeat this every time you open them – first a search, then on the result page in the speedsettings and deactivate there - so you send EVERYTHING on the private pages and every option you want to avoid - the AI monster included !
I give a you know what that its “anonymous” - I strictly refuse to send any data to anywhere most of all when its a “special privacy browser” who lures users in with this extra security and then rips them off so viciously
And this is called deception and fraud !
1 Like