LATEST START-UPS
AnoDetect
Security for Data-Science Applications
Technology
Data is getting more valuable than ever. With the ease of access to lots of information, modern algorithms are driven by data and a lot of impacting decisions are made by the outcome of these models, usually without any human interaction. Due to their importance, data streams, storage and accessing programs, which make data-based decisions, are valuable targets for attacks. Aside from classical attacks attempting to steal information, recent attacks try to manipulate existing data to change the outcome generated by the algorithms to trigger changes. These attacks are very hard to detect as the system does not show abnormal behavior. The system slowly gets poisoned by feeding it with false information which will never be detectable or only once it's too late. AnoDetect is a solution to detect and prevent these kind of attacks..
Scientific Background
AnoDetect is using state of the art anomaly detection to prevent data theft as well as malicious behavior of code or AI-Systems without affecting the applications performance.
Management
- Tobias Faber
- Andreas Hildebrandt
- André Rinau
Bitahoy
IoT & Smarthome Security
Technology
The problem of vulnerable Internet-of-Things (IoT) devices grew into a serious menace, culminating in massive DDoS attacks and an even more significant threat to the privacy of personal data. Bitahoy addresses this problem by developing a device that is connected to the home network and analyzes any network traffic. The goal is to provide an automatic solution to detect attacks on IoT devices and thus improve the security and privacy of the users.
Scientific Background
The Bitahoy Watchdog is a distributed system that can protect non-industrial IoT devices from malicious commands and data-exfiltration. After the deployment of the client-part of this system in the user’s network, it is capable of intercepting, analyzing, and filtering network traffic of all the connected IoT devices without any configuration. The system classifies devices and judges their real-time behaviors through machine learning based on previously observed benign activity. The founders of Bitahoy are graduates of the master's course ‘Entrepreneurial Cybersecurity’ at Saarland University.
Management
- Marius Bleif
- Alexander Fink
- Roman Tabachnikov
Website
ChainifyDB
Trusted Data Management Environments
Technology
chainifyDB converts well-founded scientific findings into marketable, user-oriented software solutions. These help companies to carry out transparent and unambiguously traceable transactions with different trading partners, to reduce conflict potentials and thus to save resources. For this purpose, chainifyDB uses private blockchain technology by extending existing database infrastructures with security functions. chainifyDB's cloud-based software solution seamlessly connects to the most common database management systems, drastically reducing the barriers to entry for secure digital transactions.
Scientific Background
The start-up, which is based at Saarland University, was founded in 2020 as a result of research work carried out by the Big Data Analytics Group. The two founders, Dr. Felix Martin Schuhknecht and Dr. Ankur Sharma, both received their doctorates in computer science at Saarland University. Dr. Schuhknecht's research areas included blockchain technologies, transaction processing, indexing and data management while Dr. Scharma focused on main memory data management and transaction processing in HTAP systems and private block chain systems. The third founder is Prof. Dr. Jens Dittrich. He is a Full Professor of Computer Science in the area of Databases, Data Management, and Big Data at Saarland University.
Management
- Dr. Felix Martin Schuhknecht
- Dr. Ankur Sharma
- Prof. Dr Jens Dittrich
Website
Codeshield
Security of Open-Source Software
Technology
Codeshield has developed a security tool for integrated development environments, which allows to analyze the whole software supply chain in almost real time and with high precision. With the help of static code analysis, data flow analysis and fingerprinting vulnerabilities in your own code as well as in integrated third party libraries can be detected and fixed.
Scientific Background
Dr. Johannes Späth has developed new and efficient algorithms for static code analysis in the context of his excellent dissertation and has published on this topic at international congresses. Manuel Benz has a master's degree in both computer science and IT security and has been working on the combination of static and dynamic analysis at the University of Paderborn since 2016. Andreas Dann has a master's degree in computer science with a minor in economics and has been researching static code analysis in the field of IT security at the University of Paderborn since 2016.
Management
- Dr. Johannes Späth
- Manuel Benz
- Andreas Dann
Website
Foldio
Digital Education
Technology
Children are surrounded by digital systems and are using them more than ever before. However, many of them are not aware of what is actually going on inside of all these tablets and computers and which risks arise when using digital devices. And how to prevent these risks by following a series of guidelines, from choosing the right password to dealing with cyberbullying and preventing the disclosure of sensitive information to others. Therefore, Foldio aims to prepare children for the digital world at an early age by teaching them the basics of computer science and cyber security in a playful way. To reach this goal Foldio develops educational paper-based toys accompanying mobile video games.
Scientific Background
After launching a first product version, the Foldio Starterset, a programmable paper robot that children can program to know the basics of how computers work, Foldio is currently developing a hybrid point and click adventure game that will teach children the basics of cyber security. CISPA Incubator supports Foldio professionally from the development of a learning concept to the first game prototype.
Management
- Amir Baradaran
- Michael Kellermann
Website
QuantPi
Technologies for explainability and robustness auditing of AI
Technology
Complex AI systems are inherently black-boxes with minimal insight into their internal functionality. This bears the danger of decisions that are not justifiable, legitimate, robust against external manipulations or that simply cannot be understood by stakeholders with different backgrounds. To help organizations meet this challenge, QuantPi developed an innovative framework which systematically matches questions about the functionality of AI systems with appropriate algorithms to produce relevant explanations. Furthermore, it allows users to understand how the latter algorithms work and to evaluate risks when interpreting their output.
Scientific Background
QuantPi is a spin-off of the prestigious Helmholtz Center for Information Security (CISPA), located in Germany. We develop automated and scalable solutions for explainability and robustness auditing of AI models. The QuantPi team is made up of leading researchers, engineers and business minds from premier universities around the world. Our customers include well-established companies of various industries and fastly-growing AI startups.
Management
- Dr. Antoine Gautier
- Philipp Adamidis
- Artur Suleymanov
- Lukas Bieringer
Website
Vipfy
Secure SaaS Marketplace
Technology
Companies typically use a variety of different (cloud) services in different functional areas. The administration of all these services is time-consuming, an organizational challenge and a major security risk. Therefore, VIPFY is developing a platform combining cloud services, simplifying management and providing protection against potential attackers.
Scientific Background
Vipfy developed an algorithm that can verify the users of a platform based on mouse movements. This algorithm is used to prevent external attackers from their knowledge of identification methods and data (e.g. user name and password) to be able to fend them off. This is done by analyzing the behavior of the attacker and matching an encrypted model of the right user.
Management
- Markus Müller
- Nils Vossebein
- Jannis Froese
- Pascal Clanget
Website
