Skip to main content

CommunicationPublished on 28 October 2024

Presentation of the finalists of the Cyber Start-up Challenge 2024

The three start-ups Patronus AI, QuantPi and Dynamo AI won over the jury in the selection procedure of the cyber start-up challenge 2024. They will present their innovative solutions for the «Security of artificial intelligence» on 30 October 2024 in a pitch at the Cyber-Defence Campus Conference. This conference is organised annually by the Cyber-Defence Campus of armasuisse Science and Technology (S+T).

Samuel Albrecht, specialist area Cyber Security and Data Science, competence sector armasuisse Science and Technology

AI field in the centre of the image surrounded by lines

In brief

Start-ups presented their innovative solution approaches on the topic of the «Security of artificial intelligence» to safeguard applications for artificial intelligence (AI) and to minimise potential risks. The winner will be awarded an order for a proof of concept and will have the opportunity to demonstrate their solution in a real DDPS environment.

The Cyber-Defence (CYD) Campus of armasuisse Science and Technology is closely involved in observing cyber developments and reacting appropriately. In view of the increasing cyber risks, it is important to identify and support innovative technologies early on. Start-ups have the potential to develop groundbreaking ideas which offer significant technological advantages for cyber security, even if these solutions are not yet fully developed.

As part of the Cyber Start-up Challenge, such innovations are being specifically sought to develop solutions for the Swiss Armed Forces and other security players together with the start-ups. This year, the focus is on the «Security of artificial intelligence» and start-ups from all over the world are presenting their visionary technologies on this topic. The jury eventually selected the following three finalists from the wide variety of participating companies:

QuantPi

Team QuantPi positioned sitting in a row and a row standing looking to the front

QuantPi performs a pioneering role in the area of trustworthy technologies for the development of responsible AI. Its end-to-end platform tests AI systems comprehensively for bias, durability, explainability, compliance and other critical metrics. This offers those involved in the AI life cycle a common understanding of their AI models, regardless of whether they have been created internally or have been obtained from third parties. The core of this platform is a high-performance proprietary test machine which uniformly evaluates all types of AI (LLMs (Large Language Models), computer vision, machine learning etc.). This supplies actionable insights and operationalises internal AI guidelines and regulatory conditions such as the European AI Act.

QuantPi emerged from one of the globally leading institutes for information security (CISPA). It shapes a future in which intelligent machines are deployed in a trusting and responsible manner. QuantPi enjoys the trust of some of the largest companies and institutions worldwide and remains a global leader in the promotion of trustworthy AI.

Patronus AI

Logo Patronus AI

Patronus AI is a company specialising in automated LLM Evaluation, protective measures and security. Its platform enables companies and authorities to evaluate and compare the performance of LLMs in real scenarios, to generate test cases on a large scale, to monitor hallucinations and to prevent other unexpected and insecure behaviours.

Users benefit from the use of Patronus AI by saving thousands of hours of human evaluation, accelerating the development and the trust in GenAI applications from the prototype to production and minimising the negative risks of hallucinations and prompt injections.

Dynamo AI

Logo Dynamo AI

Dynamo AI was founded from a team of Ph. D. scientists from MIT and security experts and offers a comprehensive approach to AI security and risk management. Dynamo AI is foremost in the area of compliant GenAI and enables governments to utilise secure and protected applications on a large scale.

Dynamo AI offers three products: DynamoEval evaluates GenAI models for security, hallucination, data protection and compliance risks. DynamoEnhance helps to eliminate identified GenAI risks and vulnerabilities to ensure secure and reliable operation. DynamoGuard offers real-time guardrailing for GenAI, which is adaptable in natural language and which can be executed in the cloud, as hybrid, on the spot or completely on edge devices to guarantee comprehensive data protection and security for AI. The option of executing AI guardrailing completely on edge devices with or without Internet connection offers customers from the defence sector guardrailing for critical operations.

The winner of this year’s Cyber Start-up Challenge will be selected on Wednesday, 30 October 2024.

Further information