Skip to main content
Go to the home page of the European Commission (opens in new window)
English en
CORDIS - EU research results
CORDIS

DeepKeep Safeguards AI Applications Across LLM, Vision, Spatial Sensing, Human-Machine Interaction and Multimodal Models with AI-Native Security and Trustworthiness

Project description

A new platform for safer AI models for physical applications

Large language models (LLMs) and other AI solutions are becoming common in various fields and everyday life. They offer numerous capabilities, such as vision and human-machine interaction. However, this integration also includes security vulnerabilities and data leaks that could harm their growing user base. The EIC-funded Deep Keep project aims to develop a platform for generative AI, LLMs, spatial sensing, computer vision, multimodal models, and human-machine interaction. This platform addresses challenges faced by AI developers when integrating AI into the physical world. It will ensure security and trustworthiness throughout the AI lifecycle, allow automation, adapt to edge device environments, enhance regulatory compliance, and protect against physical threats.

Objective

AI systems face many challenges, including privacy challenges, information theft, and biases, particularly in GenAI, LLM approaches, vision, spatial sensing, human-machine interaction and multimodal models. Integrating AI into physical environments poses additional barriers (e.g. security vulnerabilities, complex sensor handling, advanced deception scenarios). Companies developing AI models struggle with inadequate testing standards and high costs, while those adopting AI often lack the necessary expertise and security measures.
DeepKeep has developed a platform that can be applied to GenAI, LLM approaches, vision, spatial sensing, human-machine interaction, and multimodal models that can solve many multidisciplinary challenges for the integration of AI in the physical world.
The platform’s key features include the framework itself, providing a holistic coverage for both trustworthiness and security, addressing the entire AI lifecycle, and supporting diverse deployment scenarios; comprehensive algorithmic coverage; automatic discovery and analysis of AI model weak-segments; physical threat vector analysis and protection capabilities; AI-driven platform automation and generic applicability; adaptation to edge device environments; and regulatory compliance evaluation and enhancement.
Now is the right time for our solution, as AI is becoming an essential part of businesses and our everyday lives. In 2023, 35% of businesses adopted AI, 77% of all devices used some form of AI, 90% of leading businesses supported and invested in AI, and 85% of enterprises will expand AI with open-source models.
We founded DeepKeep Ltd. in 2021 in Tel Aviv, Israel, and target 30% of the TRiSM market with a pricing model that includes annual licenses with 3 plans (cloud plan, on-premise plan, and edge plan), and users can also purchase LLM tokens.
Our team consists of 24 highly experienced, multidisciplinary employees with 20% of our leadership team and 45% of our workforce being female.

Fields of science (EuroSciVoc)

Keywords

Programme(s)

Topic(s)

Funding Scheme

HORIZON-EIC-ACC - HORIZON EIC Accelerator

See all projects funded under this funding scheme

Call for proposal

(opens in new window) HORIZON-EIC-2024-ACCELERATOR-02

See all projects funded under this call

Coordinator

DEEPKEEP LTD
Net EU contribution
€ 2 490 250,00
Address
121 Menachem Begin
6701203 Tel Aviv
Israel
SME
Yes
Activity type
Private for-profit entities (excluding Higher or Secondary Education Establishments)
Links
Total cost
No data