AI tools woven into the big tech ecosystems, some more “open” than others, some more quiet, often require a silent sacrifice in exchange for easy-to-access technology. An alternative is open-source AI models which, much like open-source software, have a potential to give you more transparency and control.

You can study and even modify the model you are using, and if you don’t need that — at least understand how your data is processed, stored and shared. Sounds like something the viral AI photo editing apps can’t promise, yet millions keep giving them access to their private photos every day.

Join us as we explore how open-source AI benefits users and, from the perspective of our Ethical AI rating, provide examples of AI models that can help you reclaim your privacy.

What are open-source AI models?

With more people becoming aware of how data economy works, earning user’s trust becomes more challenging to the AI providers. To appeal to more conscious users, many AI vendors label their technology open or open-source. But without consensus on what open-source AI is, such open-washing could raise many risks: surveillance, information bias, effects on the labor market and creative industries, and excessive energy usage — to name a few.

In attempt to battle the potentially harmful trend, the Open Source Initiative released its official Open Source Artificial Intelligence Definition (OSAID). According to the OSAID, an Open Source AI is an AI system made available under terms and in a way that grant the freedoms to:

• Use the system for any purpose and without having to ask for permission.
• Study how the system works and inspect its components.
• Modify the system for any purpose, including to change its output.
• Share the system for others to use with or without modifications, for any purpose.

Even though the work on this definition has been endorsed by many companies and individuals, there’s no consensus in the community on whether or not to accept it in its current state, or if the open-source AI is even ready to be defined.

We too think there’s room for improvement. But with or without concrete definition for the OS AI, we can tell how four main freedoms of open-source – to use, to study, to modify, and to share – can impact life of the the AI users.

How do users benefit from open-source AI?

Open source promotes freedom and empowers users to control their software – in this case, the AI – they rely on. Here are several examples of opens-source AI systems could directly benefit you as a user:

• Proprietary AI can include hidden features like data harvesting for advertising and user profiling, embedded biases, and using your private data to refine the products without your consent. Open-source lets you make sure there are no hidden features or data backdoors – because you (and thousands, or even millions of others) can inspect the source code and understand how the software operates.
• If your software is not anymore supported by a provider, you may lose valuable data like your AI’s memory of interactions. And even though normally you can export this data, there’s no guarantee it will be compatible with alternative models. With open-source, you don’t rely on a single vendor – the software will continue to exist, and you will retain full ownership. Moreover, you are not locked in the provider’s ecosystem of applications in order to fully benefit from their AI, no-one controls your costs, your updates and maintenance.
• You can host your AI yourself or with a provider you trust, which means you fully control your data, make own decisions for your privacy, and maintain your compliance instead of relying on third-party services.

Nextcloud Ethical AI rating

At Nextcloud, we employ and integrate AI tools ourselves and we realize the risks brought by the evolving computer intelligence. That’s why we created the Ethical AI Rating that helps our users make the right decisions about what AI models to use in their work. It is more granular than the OSAID and more critical in areas like training data availability. For now, we will keep using it as it resonates with the unique needs of Nextcloud users.

Nextcloud Ethical AI Rating is based on three principles:

• Whether the training data is available and is free to use
• Whether the software (both for inferencing and training) is open-source
• Whether the trained model freely available for self-hosting

Based on how many of the criteria are met, a model receives Red , Orange , Yellow or Green rating.

Looking for open-source AI models to try? Consider these!

Nextcloud is not primarily an AI developer or vendor, so we don’t promote any AI products. Instead, we want to help our users choose the right tools to protect their privacy, and where needed, help them start using those tools through Nextcloud Hub. From the perspective of our Ethical AI Rating, here are several AI models that help you take control of your privacy.

Open-source LLMs via LocalAI

With LocalAI, you can use several open-source models like GPT-Neo or GPT-J, running them locally with ability to modify and interact with them on your server. In this case, the app also gets the Green rating in our Ethical AI rating, fully meeting the criteria.

OpenNMT

OpenNMT is a fully open-source AI translation framework available under MIT license. In Nextcloud, it can be used via LibreTranslate app that offers translation via your own server/API key.

Opus

Opus is a machine translation provider by University of Helsinki that runs locally on CPU and doesn’t let any private data leaves your servers. It meets all criteria of our rating.

GPT4All Falcon by Nomic AI

GPT4All Falcon is an LLM that combines the Falcon language model with the GPT4All interface. It is available under Apache 2.0 license and runs locally on CPU.

There are models that do not match the criteria of the Green rating, but can be worth considering since they are self-hosted and have very permissive licenses, making them mostly safe to run locally:

Self-hosted Stable Diffusion

Stable Diffusion is open-source model licensed under CreativeML Open RAIL-M License. Users have full access to the model and its code, can modify and redistribute it, given that it’s an on premise version. In Nextcloud, self-hosted Stable Diffusion can be used via LocalAI. We give it a Yellow rating, as the training data is not freely available in this case.

Whisper

In Nextcloud, you can use Whisper to enable speech-to-text conversion features. Whisper is available under the MIT License and has a Yellow rating due to limited availability of the training data.

Llama 2

Llama 2 is a free LLM by Meta that can be run on premise, and the software for training and inference of this model is open source. However, the training data is not freely available. It is offered under Llama 2 Community License.

We don’t limit our users’ choice of AI tools, and there are other models and apps that you can integrate in Nextcloud Hub and use with Nextcloud Assistant, including self-hosted models like NeuralBeagle14 7B, Smaug-72B and Meta’s Llama 3. Or famous AI like ChatGPT, DALL·E or Aleph Alpha – which are far from meeting all the criteria of our rating, but you still can easily integrate them in Nextcloud if you wish.

Build your own private AI platform – with Nextcloud Hub

Nextcloud Assistant is the first local AI assistant built into collaboration platform. It is integrated everywhere across Nextcloud Hub, helping you with your daily tasks while taking care of your privacy. Among tools supported by Nextcloud Assistant are chat with AI, text and image generation, email summaries, translation, dictation, inquiring about your own data with Context Chat, automated document creation with Context Write, and much more.

You can integrate AI models of your choice to enable those tools, creating a custom AI-powered environment according to your your privacy and security needs. And if needed, you can host your AI with one of the trusted SaaS providers in Europe, staying fully compliant with privacy regulations like GDPR and others.

Get Nextcloud Hub

AI-powered open-source collaboration platform that puts you back in control over your data.

Get Nextcloud Hub

The post How open-source AI models can help you take control of your privacy appeared first on Nextcloud.

Today, the Open Source Initiative has released its first official definition of Open Source AI. This is an important milestone. Let me explain why.

Why Open Source matters

The key tenet of open source is is that it puts the user in control. Software is ever growing in complexity and importance in our society. Software is how we do our work, how we communicate, how we pay, how we access information. When software is a black box, subject to subtle (or not so much) manipulation, users are at risk.

Risks of AI

AI brings these risks to an entirely new level. Not only because it makes decisions that are often entirely intransparent, but also because its easy, human like interface often lulls users to trust it far more than deserved.

The big AI firms have done their level best to ensure the attention of the public around risks with AI were aimed at existential, contrived notions akin to Skynet, the AI in the movie series Terminator, scenarios where AI would take over the world. In reality, the risks associated with AI are far more mundane. Surveillance, bias, explosive energy usage and job losses are the concerns we should focus on.

Need for control

And, just like with software, what matters is control. Who controls the AI, who makes decisions about what it can and can’t do, what goes in and what does not. With control, we can address the real risks of AI. Without control, we can simply hope that the billion dollar companies do the right thing.

They haven’t in the past. So we need Open Source AI. AI that that gives users the ability to study and modify the AI models that govern their lives.

Nextcloud Ethical AI Rating

Nextcloud gave this a first shot in March 2023, when we launched our Ethical AI Rating. A simple traffic light would show with green/yellow/orange/red if a given AI model was freely available, if its data was publicly available and if the code needed to run and train it was open source. This way we help users make an informed decision without restricting their choice of models and features.

Users of AI solutions deserve transparency and control, which is why we introduced our Ethical AI rating in early 2023. Now, we see big tech firms trying to hijack the term open source AI. We fully endorse the creation of a clear definition of open source AI by the community to protect users and the market.

Frank Karlitschek

CEO and fouder of Nextcloud

The wider open source community has picked up the gauntlet as well, and after extensive consultation with the community, today the OSI has announced an official definition of Open Source AI. This will help users, from private users, to governments, research institutes, hospitals and businesses, to make decisions what systems they can trust.

So, today, It is a first step in a journey and we are glad to be a part of it. Nextcloud has formally endorsed the definition, even if we think there is room for improvement. We will use it as a basis for our Ethical AI Rating. Our rating is a bit more granular and also more critical in some areas – for example, when it comes to data, we believe it should be always fully available – and thus, for now, we will keep using it, as it fits the use cases of our users more.

We look forward to your input, both on the OSI definition – on the road to a 2.0 – and on our AI rating.

The post Open Source AI is here appeared first on Nextcloud.

AI has a lot of potential — to let us do things better and faster, but also to cause a great harm to our privacy, creativity and perhaps even our mental well-being. We believe most of it is yet to be discovered, but today we need to figure out how to make it work for us, and not against us (or others).

And this is why we need AI ethics. At Nextcloud, we care about privacy and transparency, and believe that Ethical use of AI tools in both commercial and personal setting is essential.

In this article, we delve into five major challenges confronting organizations in their quest for ethical AI adoption: issues with major providers of AI tools, transparency concerns, regulatory compliance, data sovereignty challenges, and the dilemma of the single-vendor ecosystems. By exploring these challenges in a certain depth, we will try to derive knowledge and insights needed to navigate some ethical complexities of AI adoption to establish a safer and more sustainable approach to business.

Ethical AI in Nextcloud

Nextcloud Hub is an AI-powered collaboration platform that offers a freedom of choice when it comes to AI hosting, sourcing an appropriate model and choosing the right approach to AI integration.

Nextcloud makes and ongoing effort to promote the ethical use of AI tools. To assist our users, we employ our Ethical AI Rating to help them choose the right match for their limitations and principles of their business.

Learn more

Big tech and ethical use of AI

Biggest tech giants like Google and Facebook claim to scrutinize their development and use of AI, addressing issues such as bias, privacy, and accountability. Creation of special research boards, drafting ethical guidelines, and participation in various forums and high-profile collaborations help secure a position of the opinion leader and ambassador of the AI Ethics.

Those initiatives also serve as a differentiator on a competitive market and help companies improve their public image where consumer trust means everything. And unfortunately, Ethical AI adoption turns out to be a simple windowdressing where profit motives prevail over the ethical pursuits as we see big AI providers terminate their ethical teams amid growing AI product investments: 

• In early 2023, Microsoft terminated its entire AI Ethics and Society team as a part of a massive layoff, despite leading the market in the development of mass-access AI products and actively promoting ethical AI principles.
• In 2018, Google published its AI principles, committing to ethical AI development. However, in the following years the company faces a backlash over a controversial Project Maven to develop military AI for Pentagon, fires its leading AI researcher Timnit Gebru, splits up its Responsible Innovation team, and ultimately faces an internal dissent from the employees over the implementation of these principles.
• In 2023, Meta disbanded its Responsible AI (RAI) team while investing even more resources into generative artificial intelligence products.

Responsible innovation efforts and dedicated ethical teams seem like a veneer of ethical responsibility while the deeper, systemic issues inherent in AI deployment remain unchallenged.

Data collection and transparency issues

One of these challenges is with transparency of AI training practices.

Even though companies are legally required to inform users about how their data is processed, some AI providers like Meta start collecting vast amounts of data from user content, while the policy is very hard to opt out of. And this is not the only example of how tech giants cut corners to harvest data for AI training when running out of supply.

For example, in 2021, Open AI reportedly transcribed over one million hours of YouTube videos to feed data to ChatGPT. While according to two members of the privacy team at Google, in 2022 company wanted to expand the use of consumer data for AI training, including publicly available content in Google Docs, Google Sheets and related apps.

Ethical AI certification standards and regulations

Data privacy regulations play a crucial role in governing the use of AI technologies, ensuring that individuals’ privacy rights are protected and that AI-powered applications are used responsibly. From a perspective of a company employing AI tools in their business, compliance with such regulations is essential.

Data privacy regulations and the use of AI

In the European Union, such regulation is provided by the General Data protection Regulation (GDPR) and the AI Act, an embodiment of the common regulatory framework for AI. In the US, the National Institute of Standards and Technology issued the Risk Management Framework (AI RMF), which is a guidance to companies and other entities on using, designing and deploying AI. However, this framework is voluntary and does not call for  penalties for noncompliance.

In some regions, legislation also controls collection and processing of specific types of data that companies can collect unconsciously, for example protected health infiormation (PHI) and biometric data that may be collected through various AI-powered health apps. In the US, Health Insurance Portability and Accountability Act (HIPAA) and the Illinois Biometric Information Privacy Act (BIPA) aim to regulate collection of health data.

Ethical AI certification standards

Ethical AI certification standards are designed to ensure that the development and deployment of AI technologies align with societal values, promote trust, and mitigate potential harms. While noncompliance does not lead to any legal penalties, companies still face certain risks, such as reputational damage and affected stakeholder relations. Examples include IEEE Global Initiative on Ethics of Autonomous and Intelligent Systems, EU Ethics Guidelines for Trustworthy AI, ISO/IEC JTC 1/SC 42, AECP, RAIL, and more.

How to minimize the risks of noncompliance

Regulations vary by region and industry, and the first essential step is to research data protection laws relevant to your business. Noncompliance, even if unintentional, may lead to serious consequences including both fines and the reputational damage. While those regulations enforce various policies, there are common principles to bear in mind that can help minimize risks:

• Obtain explicit consent before collecting data, informing users of what data is collected and how it is used and shared. Make sure to provide straightforward ways to opt out.
• Make your data processing and storage protocols accessible to ensure compliance and ease potential user concerns.
• Use models that provide transparent information about training data and algorithms used for collecting and processing.
• Store user data and run your AI locally to avoid conflicts with your local regulations. If local hosting is not possible, vet third-party providers for their compliance with data protection regulations. Continuously monitor their practices to ensure ongoing compliance.
• Train employees on data protection regulations and company policies. Promote a culture of privacy and security awareness within the organization.
• Regularly update and patch your systems to mitigate vulnerabilities.

Nextcloud compliance kit

We provide our customers with direct consultation services and multiple resources to support their compliance efforts. This includes a high-level 12-step checklist offering an overview of key compliance requirements and a detailed administrator manual providing concrete, hands-on guidance for implementing compliance measures effectively.

Learn more

Data locality and data sovereignty

The algorithms and data usage policies used by public AI services lack transparency, which makes it difficult for organizations to ensure ethical AI practices. Besides, when employees use publicly available cloud-based AI tools as a means of work, it may cause compliance issues and privacy risks related to data location:

• Using third-party AI services may raise compliance issues with data protection regulations (e.g. GDPR, CCPA) and industry-specific standards (e.g. HIPAA, PCI DSS).
• Public AI services may store data in jurisdictions that do not comply with the company’s data sovereignty requirements, leading to legal and regulatory risks. It can work the other way round as well: for example, the US CLOUD Act allows US law enforcement to compel US-based technology companies to disclose electronic communications, regardless of where the data is stored.

Hosting AI tools and their data locally is crucial for ensuring robust data privacy and compliance. While cloud-based solutions offer flexibility and scalability, local hosting provides the control and assurance necessary for handling sensitive or regulated data effectively.

AI vendor lock-in and strategic risks to digital autonomy

AI services provided by the big tech vendors are often highly integrated with their products, providing smoother operations and better overall performance on the user side. However, the risks of being locked in the mono-provider ecosystem include little flexibility as to what tools to employ and major dependence on the vendor’s decision making and product management. That also creates strategic risks, not only for the companies, but society as a whole.

After Danish privacy regulator ruled against sharing student’s data with Google, the company reportedly promised to change the way they process data to continue supplying Google products to Denmark’s schools. That means that the schools can avoid short-term disruptions, and also save funds as companies like Google and Microsoft make their products accessible for educational organizations.

However, in the long term, by providing kids with their proprietary technology companies get a strong grip over their future choices — with kids habituated to using the products and sharing their data since childhood.

Similarly, it is important to act strategically when adopting the AI in the daily life — a negative change can happen gradually and go unnoticed until it’s very late. While today students’ data isn’t used to train the foreign big tech AI en masse, tomorrow it might be, given the steady growth of AI popularity. As kid, you will leave school and keep using Google, because it has already trained a personal AI for you and any other product will be hard to adopt.

Vendor-independence should be a part of the long-term strategy. European organizations are better off using European AI that is part of a sovereign ecosystem, hosted in Europe and trained on a local data.

Regain control of your data with Nextcloud Hub

Nextcloud Hub is the most popular self-hosted collaboration platform that integrates file sharing, document collaboration, groupware, and videoconferencing tools in a single modular interface. It is secure and private by design and gives you ultimate control over your data and ensures maximum compliance.

Realizing the big potential AI has for our daily life and work, we powered Nextcloud Hub with AI features that give you performance, but also let you care about your privacy. It features the Nextcloud Assistant, an AI-powered interface that enhances your entire platform with versatile automation features, augmented communication and content creation instruments. You build your AI stack the way you want it, with multiple apps, models and deployment formats available.

The post Ethical use of AI: 5 major challenges appeared first on Nextcloud.

Introducing Assistant 2.0

With a nearly complete overhaul of the user interface AND how our AI Assistant works on the back-end, plus a ton of new, big features, we decided that this is the moment to call this the Assistant 2.0. Let’s go over what’s new, but first a short summary of where we are today!

Nextcloud has a ton of AI features, some big and elaborate, others more basic. Some train a neural network on your data, on your own server to give you a smart inbox in Mail or warnings about unexpected logins from Suspicious Login Detection. Others rely on a pre-trained network that can recognize faces and objects in photos, generate a transcript of your video call or modify text. We can also generate images, translate text and chat messages, and much more. Many of the features are accessible through our easy to use Nextcloud Assistant interface!

Many of these features can help you, save you time, or improve your productivity. And best of all: as always we focus on making all these services available of running on your own server. Given even the US Congress doesn’t trust ChatGPT or Microsoft CoPilot, there are good reasons to want control over the data send to and from an AI service! Our Ethical AI Rating system provides you transparency about the AI used to generate your text, image or classify your photos:

• Is it open source? Can I study how it works and re-train or optimize the model for my needs?
• Is the model freely available? Can I run it on my own server?
• Is the training data open and available so I can inspect it for issues and re-train or optimize the model?

Learn more in our blog post about our approach to AI.

Context Chat and Context Write

And now, let’s explore what’s new: The two biggest new features are Context Write and Context Chat. These both share a lot in common, both providing AI-assisted outputs for you, but they’re also distinctly different.

Context Chat, in short, is the ability for the Assistant to answer questions about your data. You can ask the Assistant questions about a document you have: Say, you have a PDF with the manual of your new digital watch… you’ll be able to ask the Assistant how to set it up, or how to replace the battery. And if your company has a nicely documented reimbursement process in its documentation in Collectives, you can ask the Assistant questions about this process. This is an absolutely amazing ability!

It is not just limited to your files either: we developed an API so apps like Mail, Calendar, Talk and Deck can make their data available too. You’ll then be able to ask “When do I have my next meeting with my boss?” (from the Calendar) and “What tasks did she ask me to work on as high priority this week?” (from Mail). Or “Give me a summary of the status of project X.” (Deck + chat perhaps?)

Context Write, on the other hand, lets the Assistant rewrite something in a certain requested context, or style. Say, you have a nice poem and want to write another in that same particular style. By providing the original poem, then new text of what you want to say, the Assistant will then create a masterpiece. But you can also give it a form, give it your data, and ask it to provide the text you would need to fill in the form. Impressive, and useful!

Big user interface overhaul

Over the last year and a half, we have introduced a wide range of AI features in Nextcloud. Translation, transcription of audio, text and image generation, and more. Then the Nextcloud Assistant came – and we’ve now integrated most of our AI features into the Assistant itself, all available from a unified interface. It is no longer limited to text, but can also generate images and transcribe audio.

Another improvement many of you will appreciate is that you can now see a history, a list of earlier generated images, text or audio transcripts in the Assistant interface. You can even go back and generate more based on your earlier prompts. It is now also possible to cancel a running operation, if desired.

Lastly, as possible inputs to the Assistant you can paste or type text or record audio, and now… select one or more documents as input too!

Behind the scenes

We also did significant work behind the scenes. We made it possible for administrators to make use of AI models we’ve not yet explicitly integrated into Nextcloud, giving more flexibility and versatility to your choice of models to use.

Especially for larger deployments, we have made it possible to run the LLM and other AI operations like Transcription and Context Chat on a separate server, using our new external app ecosystem for one-click, Docker-based deployments. You can also now use a GPU for text generation, transcription and context chat. This means transcribing an hour-long audio file could, on a CPU, take many minutes while taking only seconds on a GPU!

More coming soon

And, of course, this is not all. We have developed an API which can be used by apps like Deck, Calendar or Mail to support Context Chat. This will allow the Assistant to answer questions about your mail, tasks or upcoming appointments! We look forward to these and other apps implementing support for this feature.

Nextcloud Hub 8 – our next version out very soon – will also feature two more app integrations of the Assistant: First, Nextcloud Talk will be able to show you summaries of conversation in Talk rooms, and Nextcloud Mail will introduce email ‘suggested answers’! If you haven’t yet registered for our Nextcloud Hub 8 Launch Event on April 24, this is the time to do so to find out more!

How to get it

So now, you ask: When can I get it? We have good news! Nearly all these features have been backported to Nextcloud Hub 7, meaning they are available in the Nextcloud app store right now. All you need to do is update. If you have not yet tried out the assistant, perhaps this is the time to do it!

To run the AI locally, it’s easiest to use LocalAI. A community container exists for our All-in-one, so users of the AIO can simply go to this page to learn more about adding extra containers and then enable the LocalAI container. Be sure to check out our documentation and ask questions in the forums – you could just start below this post!

IONOS, OVHcloud, plusserver and others bring you…

This brings us to the final piece of the puzzle: For these AI features to work smoothly, you will need some serious hardware. This unfortunately will not be a fun experience on your Raspberry Pi. But! Luckily, a number of businesses have started to provide AI-as-a-Service, running various open source models that Nextcloud supports in a way that you can connect to remotely. While this is great, unfortunately nearly all these companies are American-based, with a few in China. If you live in the USA, that is of course perfect – the data stays in your jurisdiction, and you have a choice of local providers. But for our users in Europe, this is less than ideal. The US jurisdiction unfortunately treats the data of European users with less than ideal respect for privacy, in no small part thanks to the CLOUD act signed into law by the former US president Donald Trump.

But today, we announce that we have been working closely with several major European hosting companies including IONOS, OVHcloud and plusserver to integrate their upcoming AI-as-a-Service solutions. This will make them available to Nextcloud customers and users who wish to take advantage of the advanced AI functionalities in Nextcloud Hub but don’t wish to host AI services themselves. Several Nextcloud customers have expressed interest in these features and are preparing to or in the process of testing them. Here’s what our partners and customers have to say:

OVHcloud is thrilled to collaborate with Nextcloud in delivering state-of-the-art Digital Sovereign AI solutions to our customers. We are aligned with Nextcloud’s commitment to offering secure, privacy-conscious AI functionalities that safeguard training data and adhere strictly to GDPR regulations. We are excited to announce that this integration will become available in Q2 2024, marking a significant milestone in our journey towards responsible AI innovation.

Germain Masse, AI Product Marketing Manager at OVHCloud

We are proud to offer Nextcloud customers a solution that complies with European data protection regulations. Reliable, trustworthy, and local AI services will be key to protecting their digital sovereignty for both the public and private sectors in Europe.

IONOS

plusKI brings cutting edge, open source AI technologies in a compliant and digitally sovereign service to the German market. We are excited to work with Nextcloud to make this available to their customers as back-end to the new generation of the Nextcloud Assistant.

Christian Schmitz, Director Open Source at plusserver

We at SUNET are happy that that Nextcloud Assistant is 100% Free and Open Source, runs on premise and provides useful features that we hope will benefit our users going forward. We enjoy the very productive collaboration with Nextcloud.

Mike Nordin, Sunet, the Swedish University Computer Network

In the future, we will make AI services available to colleagues in the state administration directly at the workplace. We want to drive the development of digitally sovereign and open solutions in addition to the use of existing AI services such as ChatGPT. Because, the administration of the future will work in an automated, algorithmized, cloudified and data-based way. To make this vision of the future a reality, we must provide the appropriate tools. This also creates value and jobs in our domestic digital economy.

Minister of Digitisation and Head of the State Chancellery Dirk Schrödter, Schleswig-Holstein, Germany

US technology firms have a large head start on European companies offering large language models and other AI services to smaller firms and end customers. There is a significant number of offerings either using their own models like OpenAI, or providing one or a choice of open models like Llama, Gemma or Mistral as a service. In Europe, Mistral and Aleph Alpha are still working on offering their own models while only a few large hosting providers are looking to offer various open LLM’s in an AI-as-a-service model.

At the same time, there is a growing demand from our customers and users for European AI services. Running AI locally is expensive, and AI-as-a-Service from foreign tech firms is not a solution for most Nextcloud customers in Europe, given the privacy and sovereignty risks inherent to shipping large amounts of crucial data overseas. For this reason, we have been working with a handful of providers to offer their AI services. To speed up the process, we connected them with potential customers. As you can imagine, offering a service when it is unclear who its customers might be is not easy. But several universities are interested in the ability to outsource some heavy AI operations, so there is now movement towards AI services. This means not only our users, but the wider European ecosystem will be able to benefit from this!

Public availability of the European AI services is expected in the course of spring and summer, with some even aiming for late April/early May!

Thank you!

We are super excited about this release and we’d like to thank everybody who contributed, with testing, reviewing code but also the universities who are willing to test the AI services, and of course OVHcloud, IONOS, plusserver and our other partners who’d like to provide these services!

The post Nextcloud releases Assistant 2.0 and pushes AI-as-a-Service appeared first on Nextcloud.

Last Wednesday, social media buzzed with apprehension over Dropbox’s new AI feature that shares data with OpenAI to perform an experimental AI-powered universal search across user data on the platform. As reported by Ars Technica and other media, the on-by-default setting led to understandable upset from many Dropbox users.

While Dropbox does offer the option to opt-out of their new feature, something that sends your data to a third party company shouldn’t be on by default without first checking with the user if they’re okay with it. At the very least, they could have enabled a pop-up dialog the first time they showed the feature, explaining the consequences and offering that opt-out in an easy way.

When we at Nextcloud introduced our AI feature set in early 2023, we recognized the challenges intrinsic to Artificial Intelligence, and from day 1 have prioritized total transparency, choice, availability to users, and the honoring of adjoining ethical implications. Nextcloud Hub features rich AI offerings with an always privacy-first approach: while many are available, our remote-AI features are optional and disabled-by-default, never exposing user data without your knowledge or consent.

For example, we released an article recently that reiterates our AI approach and explains the difference between our completely self-hosted AI options versus those implemented as integrations from external parties. We then dove into each new AI feature to illustrate their utility, the different options to enable them, and any potential AI and privacy-related implications, aided by our Ethical AI Rating system.

In Nextcloud Hub 7, we just released our advanced, Unified Search that allows you to search for any file, person and more in Nextcloud. Since the data is only indexed on your server, no data leaves your server. Nextcloud can store all your data from contacts to documents and notes, so you don’t even have a need for using or interacting with a 3^rd party service if you don’t want to.

Similarly, the Nextcloud AI Assistant uses the collective data from your server to help you enhance and speed up your work with all-local AI options. You can get language assistance in your documents, summarize your email threads and Talk chats, and more.

We just shared a tech preview of our Nextcloud AI Assistant with Context Chat, a new initiative developed together with the German state Schleswig-Holstein that enables local AI models within Nextcloud to search users’ local data across the platform. While on the surface this appear similar to Dropbox Dash – their new cross-platform search utility – the key implication and difference is in the implementation: Dropbox relies solely on sending user data to OpenAI for AI-assisted processing, while Nextcloud’s Assistant with Context Chat can rely completely on in-house, self-hosted local AI models, a real privacy [boon].

What we are illustrating here is that yes, AI can give great productivity benefits. And you can bring that to users in a responsible way. Even if you’re not willing or able to offer AI features in-house or on premise and instead want to rely on OpenAI or other big AI companies, there is no need for sharing your users’ data to such a third party without a warning.

Thank you for reading and staying up-to-date about current data privacy news that we, like you, care deeply about. We hope you find Nextcloud solves your needs and provides you unmatched opportunities to use and work with AI securely, easily, and privately. We would also love to continue the conversation: send us your thoughts via Mastodon and X (Twitter).

The post Dropbox handing data to OpenAI: the risks of AI in practice appeared first on Nextcloud.

Over the last year, AI has become a popular topic. Some is hype, some is substance. Some is good, some is bad. We want to give you the good, not the bad, and ignore the hype!

In Nextcloud Hub 4 and 5, we’ve focused on providing you the best tools that AI has to offer. Benefits like improving your work productivity and jump starting creativity, AI features can enhance your Nextcloud experience.

To fit everyone’s needs, we offer several types of integrations for AI, from completely self-hosted options to integrations with external services.

If you’re interested in setting up Nextcloud’s AI features on your instance, then keep reading for a breakdown of how to enable and use each one!

Self-hosted vs. integrations from external services

First, let’s understand the difference between self-hosted AI features versus integrations from external services.

All self-hosted AI features are built-in and only run on YOUR server, meaning no data leaves your premises and you’re in full control. Our self-hosted options can be installed from the Nextcloud App Store. Some are built-in and some require some command line commands to be run to download Ai models.

We also offer AI features using remote AI services from companies like DeepL and OpenAI (ChatGPT and DALL·E). These features can be enabled on your server but data from the app is sent to the third party. Note that nothing is ever sent to external services by default unless you explicitly enable those features or install the related apps. However, if you like these services, then they are available to you!

Ethical AI Rating

For some use cases, tools like ChatGPT might be a reasonable solution, while for other data, it is paramount to have a local, on-prem, open solution. To differentiate these, we developed an Ethical AI Rating.

For every AI feature that Nextcloud supports, a rating is established. This shows you the risks you take with your data is when using a feature and it gives you a quick insight into the ethical implications of your choice.

Now, let’s learn about Nextcloud’s AI features, their rating and how to enable and use them!

About LocalAI

LocalAI is an open source API that allows you to set up and use many AI features to run locally on your server. However instead of connecting to the OpenAI API for these, you can also connect to a self-hosted LocalAI instance with the Nextcloud LocalAI integration app.

The AI features LocalAI supports:

• Image generation
• Text generation
• Speech-to-text dictation

1. Image generation

Your cloud is full of photos, but sometimes the image you need isn’t there. Maybe it is somewhere on the web, or maybe you can only visualize it in your head? With AI image generation, you can type in keywords to create the image you’re imagining. Your instant AI generated image can be inserted into a Talk chat, Notes and much more.

3 options to generate images:

1. Online StableDiffusion based generator via Replicate’s remote API – Yellow rating
2. Self-hosted StableDiffusion using the OpenAI app- Yellow rating
3. Online DALL·E 2 – Red rating

How to enable:

To use OpenAI’s DALL·E 2 images or the on premise StableDiffusion based generator, just install the OpenAI integration app. You can find instructions to get the model in the settings!

This feature is very versatile on Nextcloud, being able to insert an AI generated image in a document, Talk chat and more. To use this feature in your Nextcloud, open up the Smart Picker by typing a slash (/) in your document, Talk chat and more!

2. Text generation

How do you craft intelligent, professional text? With the text generating tool in Nextcloud Talk, Office and Mail of course!

On-premises option: Yellow rating

• Self-hosted instance of LocalAI that runs language models locally.

External service integration: Red rating

• ChatGPT with GPT-3 and GPT-4 models by OpenAI.

How to enable:

For the on premise option, you also install the OpenAI app to communicate with a self-hosted instance of LocalAI which can run language models to generate text. See instructions.

To use OpenAI’s ChatGPT, install the OpenAI integration app.

3. Speech-to-text dictation

Sometimes, it’s easier to say something out loud. With Nextcloud’s speech-to-text dictation feature, you can formulate text through your voice. Use your spoken words to send a message in a Nextcloud Talk chat, in an email and through many more applications.

On-premises option: Green rating

• The on premise version can be enabled with the OpenAI app. In your Nextcloud settings, configure the OpenAI integration app to use your self-hosted Local AI instance.

External service integration: Yellow rating

• The external service integration option utilizes Whisper, as seen below.

4. Speech-to-text transcription

In the speech-to-text transcription feature, your call recordings can be automatically transcribed. This smart feature is great for recording what was said during important meetings for your record keeping.

How to enable: Green rating

If the OpenAI integration is installed and configured (using either OpenAI or LocalAI), Talk will be able to automatically produce speech-to-text transcriptions of call recordings. You will receive a transcription file just like the video recording file right after the recorded call ends.

5. Smart Inbox in Mail

Now let’s talk about our smart feature in Mail. Nextcloud Mail uses Machine Learning (ML) to classify the importance of incoming emails based on your historical data and pattern recognition. This provides you a priority inbox, where your most important messages are at the top. Due to the ML technology, your data never leaves the server.

How to enable: Green rating

Luckily for you, this feature is already integrated by default. However, if you want to disable it at any time, you can do so in the bottom left settings of the Mail app.

6. Face and object recognition – Green rating

In your Nextcloud Photos, you can quickly find people, places or objects when using the Recognize app. Recognize goes through your media collection and adds fitting tags which automatically categorize your photos, and even music!

The way Recognize works is through training an artificial neural network that recognizes patterns which yields the results we see. You can learn more about how the app works and facial recognition technology in the blog.

How to enable:

To get face recognition in your Nextcloud, install the Recognize app in the Nextcloud App Store.

“This app is incredible! It works really well. Very easy to set up. It takes a long time for the first recognition, but man, is it worth it! I expected face recognition working well, but I did not expect the tags feature working very well also! It created tags I did not think is possible. Super useful!”

György Kiss, Recognize apps rating

7. Background images and blurring in video calls

Before joining a video call, you have the opportunity to change your background to an image of your choice or add a blur effect. It so happens that this blur “effect” is actually a complex computing contouring that uses Machine Learning (ML). Nextcloud uses the Jitsu library and a model called MediaPipe (both open source) to create this effect.

How to enable: Green rating

This feature is ready to be used once you’ve installed Hub 5.

8. Translation tools

In today’s connected world, you’re bound to be put in a situation where you need to translate a message, word or piece of text in another language. Now you can translate anything with Nextcloud!

Local translation – Green rating

Nextcloud provides a translation service app that runs locally on your server. Using open source Opus models by University of Helsinki in Finland, you can rest assured no data can leak.

Use local translation in Nextcloud Talk to translate a message in a chat or in a Text document.

Witness how your writing can be instantly translated in Nextcloud Text.

How to enable:

Enable local translation by installing the Translate app! Languages supported are: English, German, French, Spanish and Chinese so far.

LibreTranslate – Green rating

The LibreTranslate integration provides translations using your own server/api key. This translation tool uses open source OpenNMT models.

DeepL translation – Red rating

You also have the option to use the external service integration DeepL. DeepL is a popular translator based in Germany that provides translations in 29 different languages.

DeepL receives a red rating because they use proprietary models and the app connects to their website. Since DeepL is a separate service, data put into this model on your server will be sent to them. Take care to read about DeepL’s privacy protocols and what they do with your data.

9. Suspicious login detection

Suspicious login detection detects and warns you about suspicious IPs logging into your Nextcloud.

This feature works through learning your behavior of logins – when and where – so if a suspicious login occurs, you’ll be notified in time to change your password or upgrade your security controls.

How to enable: Green rating

Suspicious login detection was once an app, but now it is integrated right into your instance. Just go to your admin settings and click on security. As always, the feature can be enabled or disabled per your preference.

The post AI in Nextcloud: what, why and how appeared first on Nextcloud.