First look at Mozilla’s Orbit AI-extension for Firefox

Posted on by Martin Brinkmann

Orbit by Mozilla is a browser extension that brings AI-functionality to Firefox. It was announced in September 2024 and is currently available as a beta.

Orbit supports a few options right now, but there are also limitations. As far as these are concerned, it is available only for pages in English at the time of writing.

Main features include:

  • Getting a quick summary of a webpage’s or a video’s content.
  • Interact with the AI in a chat-like interface.

What about privacy?

Mozilla says that Orbit does not use user data for training. It is based on Mistral 7B, which is one of the top small large language models out there.

The service does not require an account, which is good for privacy. Downside is that prompts are lost when you navigate away from a page.

Mozilla states in the privacy policy that is is getting information when the service is used.

We receive basic telemetry data by default, which we use to improve the performance, stability, and security of the Service, including information about your device, including hardware configuration and device operating system; browser information, such as browser type and settings; log files and usage data; and event information, such as errors or crashes. We may also receive additional information about your device, such as country, language, operator and OEM, which we use to prevent fraud and abuse, enforce our terms, and improve the security of the Service.

Orbit uses third-party services and shares some information with those. It uses the Google Cloud Platform as well as Sentry for application performance monitoring.

Using Orbit

First thing you do is install Orbit in Firefox. You get a privacy prompt after installation. Here you select whether you want basic telemetry to be shared only or more.

Once that is out of the way, you see the Orbit icon floating in the right corner of webpages. Hover over the icon to display the available actions. The top two, Summarize and Ask Orbit, are the main actions.

Summarize provides a short summary of webpages, emails, videos, and other content on a specific site. It is designed to return the essence of the page, but it may be too basic, especially on longer pages, emails, or videos.

The ask Orbit option is available as a standalone option and also after summaries. It works like other AI chat tools. Type a query and wait for the AI to respond.

The option works well with summaries. Start with a summary and then ask Orbit for additional details.

I tried this on several videos, and it worked really well (using transcripts). Orbit answers follow-up questions that you may have, and it worked surprisingly well in tests.

You may also chat directly with Orbit. Chats are limited to content on the active webpage. This limits Orbit.

Closing Words

Orbit, all in all, is an interesting helper extension. While it shares issues with every other AI-tool out there, mainly hallucinations, it is designed to help users make sense of content that they see in front of them.

It worked well with videos, especially with follow-up questions. Still, it may be necessary to verify the information that the AI provides, especially if it is critical.

Have you tried AI tools in recent time? If so, which is your favorite and why?

Opera

How to disable Opera browser’s splash logo on startup

Posted on by Martin Brinkmann

If you have used the Opera browser recently, you may have noticed that the browser displays a splash logo on start. It may also play sound while it is at it.

While that may look cool to some, others might feel that it slows down the start of the browser or find it outright annoying.

There is no preference in Opera to turn the splash screen off on start, at least to my knowledge.

Good news is that there is a way to disable Opera’s splash logo, but it requires manual action.

Some information:

  • Opera loads the file opera_gx_splash.exe during startup to display the logo and play the sound.
  • This file is used for all different editions of the browser, including Opera One, Opera GX, or development editions.

Here is how it is done:

  1. Load C:\Users\YOURUSERNAME\AppData\Local\Programs in File Explorer.
    • Replace YOURUSERNAME with your actual username on the system.
  2. Open the Opera folder. If you have multiple, repeat the process for each.
  3. Open the subfolder with the highest version number, e.g. 114.0.5282.233, in that folder.
  4. Right-click on opera_gx_splash.exe and select rename from the context menu.
  5. Change the name, e.g. to opera_gx_splash-DISABLED.exe.

When you start Opera the next time, the browser will load immediately. The splash screen is gone.

Note: Problem is that you need to repeat the process whenever Opera updates. If it uses a new directory, it may also create a new opera_gx_splash.exe file in it. So, whenever Opera updates, make sure you repeat the process to disable the splash screen on start of the browser.

Have you tried Opera browser recently? What is your take on it and on the splash screen in particular?

Google’s proposal for avoiding breakup of company is bad news for Mozilla

Posted on by Martin Brinkmann

You may recall that Alphabet is currently under investigation in the United States and that a potential breakup of the company is looming over the company’s head.

Suggestions include selling of Google Chrome and Android, or separating the ad business are all on the table.

The Verge reports that Alphabet has made a counteroffer. Clearly, the company does not want a splitting-up. The counteroffer does not include the splitting up of any company services or products.

Instead, Alphabet suggests the following remedies:

  • Browser companies should get more flexibility and the ability to change default providers every 12 month period.
  • Android device makers should also get more flexibility and options to preload any apps.

The first suggestion affects Mozilla and Apple for the most part. Both have search deals in place with Google that give them millions or billions each year for making Google Search the default search engine on their respective platforms.

With Google’s suggestions, Mozilla could sign search engine deals for different platforms. It might not be that beneficial to Mozilla, however, as there are not many search engine companies out there with the financial power to agree to deals.

In fact, Microsoft with its Bing search engine may be the only one that might be a potential partner.

To make matters worse, it may also give Google more of a bargaining chip when brokering deals with the organization.

For Android, device manufacturers would have more freedom to launch their devices with multiple search engines or apps. Google is enforcing certain rules right now, if device manufacturers want to include the company’s apps on their devices.

Closing Words

Google would retain control over all of its properties, if its proposed remedies would be found acceptable. Chrome or ads, for instance, are not even mentioned by Google.

What do you think of this? Should Alphabet be broken up? Feel free to leave a comment down below.

You can now call ChatGPT: what is next? Fax? Letters?

Posted on by Martin Brinkmann

So, AI systems seem to be almost everywhere already or are in the process of being integrated into pretty much any device or product.

OpenAI, maker of ChatGPT, has now announced that you may call ChatGPT to have a chat with the AI.

Call 1-800-ChatGPT, or 1-800-242-8478, to do so. You should note that you can only do so with a US or Canada number. If you live outside those two countries, you may be able to message the number via WhatsApp to at least have a chat on Meta’s messenger instead.

OpenAI has limited both options. Calls are limited to 15 minutes per month, presumably per phone number, and there is also a daily limit on WhatsApp, but that is not mentioned in the announcement.

Another restriction is that the cutoff-date is October 2023. This means that this particular version of ChatGPT does not know anything about events that happened after October 2023.

Privacy-conscious users should be aware that OpenAI stores conversations and messages, and that it may also review them. OpenAI says this is for a limited time only and only for “safety and abuse prevention purposes”. It is also only available for ages 13 and up, but it is unclear how OpenAI wants to make sure that the age limit is enforced.

Fax or Letters next?

With telephone conquered, Fax would be the next logical step for OpenAI to offer its services on. It might not be the most elegant solution, considering that it might take a while to get replies, but it would make for a great headline and do wonders for publicity.

Letters might also be an option. While it would take some serious time to get an answer or do a back and forth with the AI, it could be a great option for digital-detoxing.

Have you chatted with ChatGPT or another AI in recent time? What is your take on these systems? Useful addition or more off a plaything will little value? Feel free to leave a comment down below.

Google Search could soon analyze files you upload to answer questions

Posted on by Martin Brinkmann

Google Search has been the major search engine in most regions for a long time. While it has changed a lot under the hood in that time, it is still used by many as the main tool for searches.

AI is threatening Google’s search dominance. While it is too early to tell how good AI-based search engines will be in comparison, Google is clearly aware of the danger.

To counter this threat, Google has started to push AI into Google Search. The main user-facing feature is AI Overviews, which displays a summary based on a user’s search query at the very top of the search results.

It works similarly to how Brave Search and other search engines display AI-generated results at the top. Verdict is still out on the usefulness of the solution.

File uploads

Screenshot of Google Search with file upload. Source: Khushal Bherwani

Soon, Google users might also be able to upload files directly on Google Search to ask questions about the uploaded file. Upload a financial report, and you might ask questions about it once Google has processed the file.

This is not a new feature, considering that several AI solutions can do the same already. Even Google has a very basic file upload feature baked into search already. You can upload an image to Google Images to find out matching images online. While no AI is involved, it at least highlights that Google has the technology in place already.

X user Khushal Bherwani published an image and a short video that shows the new feature in action. It seems to work as one might expect: you activate the file upload button on Google Search, pick a file that you want analyzed, and hit the upload button.

It is likely only available for certain file types. The video shows that the feature is not complete at this point. The upload appears to have worked in the video, but Google Search did not answer the question that the user asked about the file.

Closing Words

It remains to be seen if the feature will roll out to more users in the coming months or if it gets pulled before release. There is also the question of its usefulness and whether users feel comfortable uploading files to Google.

What is your take on the feature? Would you upload files to Google to get the chance to ask an AI question about it? Feel free to leave a comment down below.

Windows 11 Search

Windows 11: Microsoft separating Windows Search from Web Search

Posted on by Martin Brinkmann

One of the first things that I disable on Windows 11 PCs is the integrated web search. Microsoft merged local and web search in Windows some years ago mostly for its own benefit.

Web search results generate clicks and revenue for Microsoft. For users, they serve little purpose. Often, they tend to get in the way.

In best case, search returns a matching result that the user can click on. The same can be achieved in any web browser with the added benefit of having extensions and access to numerous search engines.

Microsoft may split local and web search

It seems that Microsoft could soon split the two searches in Windows 11.

https://twitter.com/alex290292/status/1866954297369563644

The feature appears to be in testing right now in development builds. With it activated, Windows moves web-based searches into a separate tab in the search interface.

This works similarly to how Windows Search separates different types of search data already. You can switch to content-specific tabs, such as Apps or Documents, to get them returned only.

The change is helpful, as it moves web-based results out of the way and restores the focus on local results.

Deskmodder has tested the feature and published several screenshots of how it looks like.

It should help users who do not know how to modify the settings to disable web search, if it bothers them. Experienced users may still want to turn off web search, provided that they have no use for it.

In case you want to give it a try, do the following:

  • Download and unpack the latest version of ViVeTool in a development build of Windows 11.
  • Run the command from an elevated Terminal prompt: .\vivetool /enable /id:54865932
  • Restart the PC.

Note that you need to run it from the ViVeTool directory. Bing Search / App need to be enabled for the change to become visible on the system.

Do you use Windows Search to search the web? Or do you ignore or disable the feature?

Copilot key. Source: Microsoft

Microsoft suggests to remap the dedicated Copilot key, but only for one user group

Posted on by Martin Brinkmann

Microsoft introduced the new dedicated Copilot-key for Windows keyboards less than a year ago. Now, the company suggests to remap the key to another function, but only on managed PCs.

Here is a quick refresher: Microsoft announced the Copilot key in January to “kick off the year of AI-powered Windows PCs”. The company even went so far to say that it might have the same impact as the introduction of the Windows key 30 years ago.

A tap on the key opens Copilot. Microsoft did add options to remap the key in Windows 11.

A new blog post on Microsoft’s Windows IT Pro Blog suggests that organizations remap the Copilot key for their users. Instead of launching Copilot, Microsoft suggests to launch the Microsoft 365 app instead, as it gives easier access “to Copilot experiences designed for work and education”.

New PCs and newly provisioned PCs will enable the new mapping of the key automatically. When a user presses the key, Copilot within the Microsoft 365 app opens.

The behavior is different for existing PCs. Microsoft says that the original mapping remains on these devices. In other words, it will open the Copilot app when pressed. This is the default for users who sign in with a Microsoft account. Users who sign in with an Entra account will be redirected to the Microsoft Copilot experience on the web instead.

Microsoft recommends that Enterprise customers configure policies to remap the Copilot key to launch the Microsoft 365 app. The app is preinstalled on all Windows 11 PCs according to Microsoft.

Microsoft even recommends uninstalling the Copilot app to “avoid confusion for employees”.

Closing Words

How useful is the Copilot key? Compared to the Windows-key, it offers little use. Press it and Copilot opens. You can achieve the same with a click on the Copilot icon.

Now, Microsoft is already suggesting to the Enterprise to remap it. It is doubtful that the Copilot key will be a smash hit going forward, unless Microsoft starts to add more features to it.

Now it is your turn. Copilot key yay or nay? Is it useful, at least for users who use Copilot regularly on Windows? Feel free to leave a comment down below.

Suggested what? Microsoft deprecates suggested actions in Windows

Posted on by Martin Brinkmann

Remember suggested actions? Microsoft introduced the feature in mid-2022 in Insider builds of Windows 11 and rolled it out to stable systems later on.

The core idea behind the feature was simple: highlight certain text in Windows and the operating system suggests actions based on the highlighted string. Highlight a phone number and you get a call option. Select date or time, and you get an option to add an event to a calendar.

That is all there was to it and Microsoft did not improve the feature since its introduction. Microsoft never added support for additional types of data. No email or physical addresses to write to or look up, no chat names, or anything else that might come to mind.

Suggested actions are out

Today, Microsoft added suggested actions to the list of deprecated features. The company writes:

Suggested actions that appear when you copy a phone number or future date in Windows 11 are deprecated and will be removed in a future Windows 11 update.

Deprecated features remain in Windows for the time being but will be removed eventually. Features that are put on the list do not receive major updates anymore and it is very rare for a feature to be taken off the list.

In other words: expect suggested actions to be removed in a future update for Windows 11.

Is it a big loss? I disabled the feature on all systems and do not know anyone who found it useful. Yes, chance is that some users found it useful.

Considering that it saves you a few clicks at most, provided that the programs that you use to call or plan events are supported, it is probably a minority of users that will miss it.

Tip: you can check out this guide on Ghacks to find out how to turn off suggested actions in Windows.

Closing Words

When Microsoft announced suggested actions, I did not really see a big use case for it. Phone names and date / time strings support limited it from the get-go and it saved users just a few clicks in the best case. Usage must have been low, as Microsoft never improved the original feature.

Now it is your turn. Did you use suggested actions? Or did you disable it or never even realize that it was there in first place?

Mozilla removes Do Not Track from Firefox and suggests alternative, but there is a better one

Posted on by Martin Brinkmann

Mozilla plans to remove the Do Not Track feature from Firefox. The idea behind it was simple: inform websites that the user of the browser does not want to be tracked.

What looked good on paper did not work well in the real world. Many sites ignored the header, which made it ineffective as a privacy tool.

Related:

Mozilla removes Adjust marketing integration from Firefox Mobile

Mozilla confirmed the removal of Do Not Track on its bug tracking website.

Global Privacy Control is the alternative

Global Privacy Control was created by several companies in 2020 as a successor to Do Not Track. The core difference to Do Not Track is that it is designed to be mandatory instead of optional, at least in some regions where consumer laws are in place.

Firefox users may enable the feature in the following way:

  1. Select the Menu button and then Settings.
  2. Switch to Privacy & Security.
  3. Check “Tell websites not to sell or share my data” under Website Privacy Preferences.

Is there a better alternative?

Whether advertisers, Internet sites, marketing companies, or other companies and services that track users honor the new Global Privacy Control feature is not in the control of the individual user.

Yes, some companies may get sued if they do not, but there is a good chance that this won’t reach mass adoption in the coming years and that tracking continues to take place.

That leaves taking care of tracking as good as you can by yourself. In fact, installing a content blocker and disabling third-party cookies are two of the best options in that regard.

While you could do more, these methods alone will block the bulk of tracking that you would otherwise be subject to on today’s Internet.

So, pick uBlock Origin and install it in a browser that is not operated by a multi-billion Dollar company. Then, open the Settings of the browser and disable third-party cookies.

Note: in some rare, very rare, instances, third-party cookies may prevent functionality on a low number of websites. If that is the case, you may still set exceptions for these sites while keeping third-party cookies blocked for every other site.

Now it is your turn. Do you enable privacy features such as Do Not Track or Global Privacy Control? What do you to block tracking on the Internet?

Ice

Copilot Vision monitors your online activity for your own benefit

Posted on by Martin Brinkmann

Copilot Vision is an experimental AI-feature that, Microsoft promises, is going to revolutionize the online life of computer users.

Microsoft says that Copilot Vision sees the pages that users are on, reads the content of the pages, and is there for users when they have questions about current or past content, or want to have a discussion about it.

Browsing is no longer a “lonely experience with just you and all your tabs” according to Microsoft. How awfully nice.

Good news is that Copilot Vision is an opt-in experience. It is only available for select Pro subscribers at the time of writing and can be enabled under Copilot Labs. Furthermore, it will only work “on a select set of websites initially”.

Microsoft has come up with examples to demonstrate the usefulness of Copilot Vision.

  • Use it to plan a day at the museum, by “pointing out all the information you need to know before you visit”.
  • Tell you which products on a page match your needs and preferences.
  • Help you learn new games, for instance Geoguessr.

And privacy?

Copilot Vision is opt-in, which means that users have to enable it before it starts monitoring activity.

Microsoft says that Copilot Vision data is only kept during sessions and deleted afterwards. This means that everything a user said during a session and the context is deleted. Copilots responses are logged, however, according to Microsoft to “improve safety systems”.

Closing Words

Is Copilot Vision a useful tool? Who is it for? The examples that Microsoft provided do not sound overly spectacular. Telling me which products I most likely like on a page? Maybe on a page with thousands or products and endless scrolling for a preselection, but otherwise?

There is also the question of trust.

  • Do you want an AI to monitor your browsing, even if it is just for a session?
  • Do you trust Microsoft to delete the data after the session?
  • Do you believe that the answers that Microsoft logs do not contain personal information?
  • That Copilot Vision won’t get enabled automatically, by error.

Now it is your turn. Could Copilot Vision be a helpful tool in the future? Or do you see issues and problems that it could cause first and foremost? Feel free to leave a comment down below.