Project funded in the call for proposals will address the challenges in cybersecurity and reliable AI that arise from the introduction of advanced digital solutions. results of the projects will strengthen the industry's ability to create robust and secure systems that can handle today's and tomorrow's threats. With digitalization comes great opportunities, but interconnected technologies, such as cloud services, AI and the Internet of Things, also entail increased risks. Cybersecurity is therefore crucial for Sweden's industrial competitiveness and societal security.
The call for proposals has two focuses:
- AI as an enabler for cybersecurity. Project within this focus will use AI to prevent, detect and manage cyber threats.
- Reliable and secure Project within this focus will develop robust and secure AI solutions that can withstand attacks and disruptions.
1. AI as an enabler for cybersecurity
Underlying needs
Swedish industry needs AI-based solutions that strengthen cybersecurity and automate the management of secure complex data flows. The need is increasing in line with the rising number of sophisticated attacks and the increasing dependence on digital systems. AI technology has potential to play a central role in increasing the security of complex digital systems.
With innovation and increased use of advanced AI tools the call for proposals will stimulate the development of technical solutions for automated monitoring and risk management. It will also pave the way for organizational models for efficient and secure processes. The focus will be on self-learning security that enables seamless authentication, authorization control and incident response – without human intervention.
Expected effects
The projects funded under this direction will lead the way towards next-generation cybersecurity. They will, through AI technology and innovative methods, develop solutions for advanced threat and risk analysis, automated discovery and autonomous security to protect against threats and prepare Sweden for the challenges of the future.
In the short term, the project results:
- Harness the potential of new technologies, such as AI, automation, and advanced analytics tools, to streamline cybersecurity efforts and raise the level of protection.
- Reduce the risk of disruptions in digital systems by identifying and addressing vulnerabilities before they lead to incidents.
- Improve the management of disruption consequences through faster response and recovery to minimize the impact on operations.
- Reduce human intervention in critical digital processes using self-learning security solutions such as automated authentication, authorization control, and incident response.
- Stimulate innovation in cybersecurity by promoting collaboration between industry, academia and technology providers.
In the long term, the project results:
- Strengthen Sweden's ability to develop robust and secure digital solutions to increase competitiveness and societal security.
- Create synergies between AI and cybersecurity and develop and integrate AI and cybersecurity as collaborative technology areas.
- Increase access to cybersecurity expertise for industry and business, and create a skills base that meets the needs of the future.
2. Reliable and secure AI
Underlying needs
The rapid growth of AI in various applications creates a fundamental need for reliability, security and control, for example in complex environments where multiple AI agents interact. To achieve this, both technical and organizational solutions are required to ensure that AI is implemented and used in a safe and responsible manner.
Reliable and secure AI requires robust agent systems, methods for validating and certifying the security of models, and techniques that enable structuring and specializing one's own data for secure use of large, general models.
To implement AI safely, and ensure protection against AI-based attacks and system failures, the development of new security solutions and methods, such as threat and risk analyses, is required.
Some key needs are:
- Protect training data to ensure integrity and prevent sensitive information from being recreated from generative models, while protecting the created models from unauthorized access and manipulation.
- Traceability and robustness in data sources, with the ability to verify origin and quality and derive conclusions from secure data sources .
- Measure accuracy, robustness, and confidence in responses from AI models to ensure accurate and predictable results.
- Evaluate AI systems' predictions and decisions, especially as systems become more autonomous, increasing the need for control and protection around decision-making.
- Methods for validating and certifying how robust and secure systems are.
- Error control in composite AI systems, where classical AI, generative AI, and traditional hardware or software interact. It also includes understanding and managing the emergence and propagation of errors.
- Counteract bias in data using AI to reduce the risk of unsafe or unwanted system behavior.
Expected effects
The projects funded under this direction will enable reliable and secure AI use in industry. By reducing the reliance on methods that do not meet eligibility requirements for security, traceability and trust, the implementation of AI can be accelerated in a responsible manner. The results are also expected to contribute to the business community gaining access to expertise and innovative solutions that raise the general level of security.
In the short term, the project results:
- Increase the industry's adoption of safe and reliable AI by integrating security aspects from the start , and thereby reducing dependence on unsafe methods.
- Ensure industry access to advanced cybersecurity solutions that can be implemented in existing and future AI systems.
- Minimize the risk of disruption in AI use and manage the consequences of disruptions better.
- Accelerate responsible AI implementation with traceability, trust, and robust solutions.
- Stimulate continued development of existing cutting-edge expertise in line with the needs owners' AI development .
- Development of cybersecurity solutions from technical and organizational perspectives
In the long term, the project results should lead to:
- Increased digital resilience through secure and robust data flows in industry and critical infrastructure.
- Higher efficiency when industry confidently implements new technology.
- An innovative ecosystem for digital resilience and cybersecurity that promotes knowledge transfer and skills dissemination.
- Strengthened capacity of Swedish industrial companies and research environments to develop secure digital technology through targeted project and collaboration.