AI for the future of school education 

In this article you will read about,

• how schools can benefit from artificial intelligence,
• what requirements AI must meet in the classroom, and
• why DeutschlandGPT chose the Open Telekom Cloud.

Artificial intelligence is widely recognized by the German public as a key technology of the future. One thing is clear: without AI, Germany risks falling behind in global competition. Nevertheless, one in three companies in Germany deliberately avoids using AI – primarily due to concerns about data protection, data security, and data sovereignty.

A core challenge is that the leading AI providers are based in the US or China and typically process data on servers located abroad. This raises concerns that sensitive information could be intercepted, stored, or accessed by foreign entities. Despite these risks, many organizations continue to use these solutions due to the significant gains in productivity and efficiency they offer.

DeutschlandGPT aims to position itself as a homegrown alternative to the dominant US and Chinese AI providers. This is artificial intelligence made in Germany, for Germany. The software developers, based in Germering, Bavaria, specialize in creating a sovereign AI platform. “Our mission is to create secure and independent AI solutions that unleash creativity, promote innovation and offer sustainable solutions to current challenges,” says Leonhard Benkert, CEO of DeutschlandGPT.

AI from DeutschlandGPT stands for sovereignty, security, data protection, transparency, and expert support, and is certified according to ISO 27001. The company’s AI solutions combine the power of modern AI with a strong commitment to protecting digital autonomy. This makes their AI trustworthy and sovereign. In addition to businesses, DeutschlandGPT also serves private individuals and public sector clients.

Germany’s education system is undergoing significant changes. Nearly every school is facing challenges such as unfilled positions, teacher shortages, and canceled classes. At the same time, educational requirements are continually evolving. Teachers must adapt to new learning models, teach emerging skill sets, and integrate modern media into their instruction. The potential of artificial intelligence to support the future development of education is clear. AI can effectively assist with classroom teaching and school projects, easing the workload for educators. Students gain early exposure to professional uses of modern technology, enhancing both their digital literacy and overall education.

With its current project “BildungsLLM” (LLM meaning Large Language Model), DeutschlandGPT is returning to the educational roots of its parent company, Titanom, which began as a software developer for the education sector. “Our aim is to combine our many years of experience in education with our expertise in AI development and provide learners in Germany with a high-quality didactic language model,” says Leonhard Benkert, CEO of DeutschlandGPT.

Common LLMs, such as GPT-4, are well-developed, widely accepted, and frequently used in business settings. However, several challenges prevent their official use in schools. One key requirement is that these systems must be hosted within the European Union. Additionally, there must be transparency about how the model provider handles user data, which is typically not the case with publicly available models from the US.

These concerns are justified. Student inputs may be stored outside of Germany, and tracking or identification of students is possible. Furthermore, simply providing answers does not align with educational goals. Schools require a “didactic” LLM – one that helps students develop their own solutions and reach learning objectives. Rather than supplying ready-made answers, as is common in business contexts, the model should guide learners through the process of discovery and problem-solving. The model must also offer legal certainty for schools and include safeguard functions that ensure inputs and outputs follow ethical standards, security requirements, and content restrictions. Harmful or inappropriate requests must be blocked, and responses must be filtered accordingly.

For DeutschlandGPT, another critical question arose: where should the final model be securely and legally operated once completed? The answer came from Telekom, with its Open Telekom Cloud providing a sovereign and legally compliant hosting solution.

At the beginning of 2025, DeutschlandGPT developed an educational LLM specifically designed for use in German schools. This “didactic” base model is not only tailored to the pedagogical requirements of the classroom but is also optimized for the German language.

The model is built on an advanced version of the open-source Llama 3.3 70B architecture. For this development, DeutschlandGPT used a combination of proprietary and synthetic data, which was directly integrated into the new model. To ensure student privacy, interactions with the educational LLM are not stored. In addition, the model includes integrated guard functions that enforce ethical use. These safeguards are a core security component of the system. They serve as a first line of defense against abusive or dangerous inputs by detecting potentially harmful content before it is processed and automatically blocking it. These functions not only protect the integrity of the model but, more importantly, protect students from inappropriate or misleading responses.

Since May 2025, the educational LLM has been offered as part of the AI Foundation Services from the Open Telekom Cloud. The Open Telekom Cloud provides DeutschlandGPT and its customers with a secure, sovereign, and scalable operating environment, hosted in Telekom’s highly secure data centers in Magdeburg and Biere. As a result, the BildungsLLM complies fully with all legal requirements. It is easily accessible via API and can be integrated into existing learning platforms without difficulty. The outcome is an ideal solution for secure, next-generation education.

With the Open Telekom Cloud, DeutschlandGPT has found the ideal operating platform to meet the needs of both the company and its customers in the education sector. It provides robust sovereignty, data protection, security, and scalability.
The cloud infrastructure enabled a fast launch of the service in the German market. Thanks to a usage-based billing model, the Open Telekom Cloud also supports DeutschlandGPT’s business growth by aligning costs with revenue. As an independent software vendor (ISV), DeutschlandGPT does not need to manage the infrastructure itself and can focus entirely on the continued development and marketing of its pioneering AI service.

By running in Deutsche Telekom’s sustainable data centers, the carbon footprint of the AI service is significantly reduced. These data centers are powered entirely by renewable energy and have received multiple certifications and awards for their sustainability, including the Platinum Award for green data centers. This ensures that the AI model is delivered in an environmentally responsible way.

With the educational LLM, schools gain access to a ready-to-use tool that elevates teaching and learning while remaining legally compliant and secure. The potential of AI is effectively unlocked for the future of school education.