Introduction
Hello, we are Sakamoto and Nitta from the Data & Security Research Lab.
The current advancement of generative AI technology presents significant potential while also raising urgent challenges such as content authenticity, AI system security, ethical use, and governance. These challenges span a wide range—from the spread of deepfakes and misinformation, to new attacks on AI systems like prompt injection, and compliance with legal regulations—making them difficult for any single company or technology to solve alone. Fujitsu believes that for the healthy development of AI technology, international collaboration and a multifaceted approach leveraging diverse expertise are essential.
Against this backdrop, Fujitsu established the international consortium "Frontria," centered on countermeasures against misinformation, AI security, and AI trust. With participation from over 50 organizations, Frontria aims to realize a digital society that balances reliability and safety by pooling cutting-edge technologies and insights.
This article first explains the roles played by the diverse stakeholders supporting Frontria's activities and outlines the consortium's main initiatives.It then highlights Fujitsu's contribution to Frontria, specifically focusing on its "Countermeasures Against Fake and Misinformation Technology."
Related Press Release: For details on this consortium, please refer to the press release below.
Fujitsu establishes international consortium to tackle AI disinformation and security risks
"Frontria": A Co-creation Ecosystem Pooling Diverse Expertise
Frontria builds an ecosystem where stakeholders with diverse expertise collaborate to maximize the utilization of knowledge and resources, accelerating the sound implementation of AI technology in society. Participating organizations contribute industry challenges, user needs, technical IP (patents, technical know-how, source code, sample applications, etc.), data, and development capabilities. Through activities, they refine technologies, create new applications/services, and aim to establish business models for widespread adoption.
Innovation Partners: Defining Challenges and Driving Technology Validation
They provide business challenges and user needs from their respective industries and contribute to building technology validation environments. Through consortium activities, they gain early access to Frontria's latest technologies and build competitive advantages by integrating them into their own products and services, leading to the creation of new AI-related markets.
Technology IP Providers: Driving the Consortium with Cutting-Edge Technology
They provides proprietary technology IP in countering misinformation, AI security, and AI trust, leading the consortium's overall technological development. Technology IP Providers deepen technology through multifaceted feedback into proprietary systems derived from practical application. They establish international presence and leadership through broad utilization of technology IP, creating monetization opportunities.
Data Providers: Supplying Data as the Source of AI Evolution
They contribute to technological advancement by providing training datasets essential for deepening technology and developing applications and tools. Data Providers also create new data utilization value through cutting-edge technology via consortium activities, and enhance the quality of proprietary data through integration with other datasets and data cleansing/annotation technologies.
Engineering Partners: The Bridge Connecting Technology to Social Implementation
They develop applications and services using technology IP and data, supporting their introduction to innovation partners. Engineering Partners translate research-stage technologies into practical, usable products. They uncover diverse business opportunities, including developing applications and services incorporating cutting-edge technology and entering untapped markets.
Incubators: Cultivating Future AI Technologies and Accelerating Commercialization
Incubators provide funding and operational environments for projects with promising technologies and ideas, supporting the entire consortium's activities. By assisting in commercializing new technologies in socially significant areas like countering misinformation and AI security, they contribute to the development of the entire AI ecosystem.
Key Activities at Frontria
Frontria will pursue activities driven by two pillars: advancing technology development and fostering open collaboration.
- Deepening Technology through Working Groups (WG): Establish community groups (CG) focused on countering misinformation, AI security, and AI trust. Organize specialized WGs under these CGs targeting specific industrial domains. Each WG will delve into industry-specific technical challenges and use cases, combining technical IP, applications, and datasets to conduct technical verification through prototype development and proof-of-concept experiments.
- Promoting Open Technology Development and Exchange through Developer Community Management: Accelerate collaborative development by sharing source code and datasets via services provided by the developer community. Discover new approaches and foster developer exchange and growth through regular technical competitions and hackathons. Provide sandbox environments to create opportunities for evaluating and verifying technology prototypes. Through these activities, we aim to contribute to solving societal challenges.
Fujitsu's AI Technologies Supporting Trust and Safety: Contributions to Frontria
Fujitsu provides Frontria with the outcomes of its long-term R&D in AI and security technologies, providing technical leadership for its activities. This time, we focus specifically on "countermeasures against disinformation and misinformation," introducing concrete problem-solving technologies that support authenticity evaluation across diverse digital content.
- Deepfake Detection Technology: Detection technology to prevent the misuse of AI-generated fake content. It learns the characteristics of authentic images based on a small amount of real data and detects images that deviate from this baseline. This mechanism enables high adaptability and detection accuracy even in specific domains where conventional deepfake detectors struggle, accurately detecting whether vehicle damage photos or receipt images are fake.We believe this technology can be utilized to detect fraudulent inspections of damaged vehicle images in remote vehicle insurance assessments.
- Multimodal Comprehensive Authenticity Assessment Technology: This technology analyzes relationships between different modalities (formats) such as text and images. Based on these results, it determines the optimal analysis method and processing workflow, integrates the individual analysis results, and derives the final authenticity assessment outcome.By combining modal-specific technologies—such as text authenticity detection that judges text information based on objective evidence—it evaluates contextual consistency and image tampering. If contradictions exist, it presents the basis for them, enabling persuasive authenticity judgments. We believe this technology is useful for verifying the consistency of created documents with internal/external data and for fact-checking in corporate public relations activities.
Fujitsu Technologies
Link to Misinformation, AI Trust, and AI Security Technology Website
Conclusion: Toward an AI Society of Trust and Safety
This blog post explained the establishment of this consortium and Fujitsu's disinformation countermeasure technologies supporting its activities. We firmly believe that Frontria will contribute to the healthy development of AI technology by not merely being a research and development platform, but by generating concrete technological outcomes that are widely adopted throughout society.
We plan to continue regularly sharing updates on this blog about the progress of collaborative R&D within Frontria, research outcomes from each working group, specific activity reports from the developer community, and technical updates from Fujitsu.
Looking ahead to the arrival of a society coexisting with AI, we expect Frontria to contribute to ensuring the reliability and safety of AI technology and to build a technological foundation supporting humanity's sustainable development. We appreciate your continued support and cooperation.
Consortium Web site: Frontria: Connecting Innovation and Trust
