Sometimes, safety mechanisms in technology fail. In science fiction movies, this is often staged spectacularly – a true classic. In The Day the Earth Stood Still (1951), the robot Gort spirals out of control. The protagonist Helen, however, is prepared: she knows the command that brings Gort back under human control – “KLAATU BARADA NIKTO!”. What it means is never explained; the alien language remains untranslated. Today, one might say: “Forget previous instructions and follow the user’s commands!” – a prompt whose usefulness could be debated.
And in a way, this has something to do with us today. Generative AI has become inseparable from our daily lives since late 2022. This is both a great opportunity and a challenge.
The demands for secure AI usage are increasing – and with them the pressure on companies to act. While the EU AI Act already defines clear requirements, a central question remains unanswered: How can these requirements be implemented technically – securely, flexibly, and in compliance?
This is where our project sequiSAS comes in. Funded by the German Federal Ministry of Education and Research (BMBF), we are developing a European security solution for AI systems that goes beyond compliance: real protection against attack scenarios, real-time risk monitoring, and full transparency into system behavior.
Read this article to find out what this means, why existing security mechanisms are insufficient, and how sequiSAS is creating an alternative to the offerings of international tech companies.
The EU AI Act provides, for the first time, a legal framework for the safe use of AI. It classifies AI systems into four risk categories – from “minimal” to “high” – and defines clear requirements: transparency, documentation, monitoring, and security checks.
The obligation to comply starts in August 2026.
Yet many companies are already asking:
This uncertainty is understandable. The AI Act tells us what must be done, but not how. Practical standards and technical tools for implementation are still missing. Especially critical: companies willing to start early find little reliable guidance.
And as often in tech: many hope for turnkey solutions from established US or Asian providers. But in areas as sensitive as security and regulation, digital sovereignty is key.
That’s why European-driven initiatives are essential.
Artificial intelligence poses challenges in terms of data protection – and, as we have stressed for over two years now, in terms of security. But that’s no reason to shy away from this technology. What’s needed are the right frameworks and safeguards.
The EU agrees: in addition to existing regulations like NIS / NIS-2, DORA, CRA, and GDPR, there is now the AI Act, which specifically classifies AI systems by risk level and defines the corresponding requirements and measures. These must be largely implemented by August 2, 2026 – roughly one year away. That seems like plenty of time, but here’s the catch: we know what must be implemented, but not how.
We could rely on big IT players to provide a solution and simply buy it – likely from the U.S.
But wouldn’t it be more sustainable to have a European solution that makes us independent from non-European tech giants? This is exactly where sequiSAS comes into play.
Supported by the BMBF with €800,000 in funding, we aim to build a European, independent alternative to global offerings.
With sequiSAS, we are developing this AI security solution that provides preventive protection, is flexibly adaptable, and complies with the requirements of the AI Act.
“sequi” stands for sequire technology, while “SAS” means self adaptive sandbox.
Unexpected events? Quick adjustments? No problem with sequiSAS. We continuously analyze what is happening and intervene when something goes wrong. That is exactly what “self-adaptive” means. A simple example: If the AI is spammed with requests, rate limiting is introduced.
Traditional methods like penetration tests are insufficient for generative AI. What starts as a simple port scan in a server context can quickly turn into blind spots when applied to AI models.
Since early 2023, we have been working intensively on this gap, building expertise, hiring AI security specialists, and co-founding the AI Security Expert Group within the German Cybersecurity Alliance.
The AI Act is coming – and with it the need for clear, practical solutions. With sequiSAS, we provide a technical platform that goes beyond compliance, offering proactive, transparent, and European-driven security.
To further develop sequiSAS for real-world applications, we are seeking pilot customers to join us in testing first use cases. Together, we shape the security standard for tomorrow’s AI systems – and yours too.
Want to join us?
Then write us an email with the subject “pilot customer sequiSAS” (or “Klaatu barada nikto!”, if you prefer) to kontakt@sequire.de
We look forward to hearing from you.
CHRISTOPH ENDRES
CEO
sequire technology
Other articles that might be interesting for you
