Skip to content

Home

Digital Anti-Fake Technology - Trustable Internet
Can you really trust the information on the internet?
Aiming to realize a "Trustable Internet" where you can discern truth and use it with confidence.

The Social Problem of Digital Fake Information Circulation

Image of Fake Information Circulation
Fake information generated by AI is a significant problem online. Checking facts takes time and is hard. During disasters, wrong info is hazardous. We need good ways to check information to keep people safe.

Challenge in Trustable Internet

Image of Realizing Trustable Internet
Fujitsu's anti-fake technology looks at text, pictures, sounds, and video. It detects real information, AI-generated information, and mixed information. It helps people make quick, accurate decisions, especially in emergencies such as disasters.

The benefits of Digital Anti-Fake Technology

  1. Discovery of Fake Information
    		2.
    • Information from various media sources, such as social media posts, news articles, and videos, can be analyzed. It detects lies and manipulation from information in any format, including text, images, and audio.
  2. Improvement of Information Reliability
    		3.
    • Adding credibility to uncertain information supports the judgment of internet users and contributes to building a reliable information environment.
  3. Improvement of Work Efficiency
    		4.
    • Automating fact-checking significantly reduces the time required for work by companies and individuals.
  4. Support for Quick Decision Making
    		5.
    • Supports quick decision-making based on accurate information in situations where swift judgment is necessary, such as during disasters.
  5. Suppression of Misinformation Spread
    		6.
    • Contributes to maintaining a healthy information environment for society by preventing the spread of misinformation on social media.

Technology Overview

Target Industry/Users

【Industry/Type】Public institutions (municipalities, etc.), private companies (news media, fact-checking organizations)

【Service】DRPS (Digital Risk Protection Services)

Challenges in Target Industry and Operations

"Realistic fake news" and "truth that seems false" are mixed, making it difficult to judge the correctness of internet information.

Technical Challenges

It is not easy to judge whether internet information is correct based solely on the data itself.

Solutions

Collect third-party information and evaluations that serve as evidence for the information you want to verify, and judge authenticity through "collective intelligence."

Fujitsu's Technological Advantage

Text Fact-Checking

Text Fact-Checking is the task of determining whether the content of a given text matches the facts. It includes fact-checking, evaluating the reliability of sources, and detecting logical contradictions, and requires advanced language understanding and knowledge. While adopting the standard processing framework used in conventional automatic fact-checking technology, we enhance judgment accuracy with our proprietary analysis technology.

Evidence Graph (Endorsement Graph)

It can manage complex information relationships by associating evidence information/evaluation with the person/organization/AI that issued it. The absence of contradictions in the evidence, when presented as a graph, proves the correctness of the information and enhances the reliability of the authenticity judgment.

By incorporating the above technologies, we plan to develop the world's first countermeasure platform for fake information by the end of fiscal year 2025, which integrates detection, evidence collection, analysis, and evaluation of fake information.

The benefits of Trustable Internet (Details)

  • Prepare for threats lurking nearby, such as deepfakes and fake news, by verifying the credibility of internet information and supplementing it with human judgment.
  • Collect, analyze, and evaluate evidence to enhance the credibility of uncertain information.

Use case

Public Institutions (Municipalities, etc.)

  • Fact-checking during disasters
    • Used for fact-checking fake information during natural disasters such as floods, landslides, and earthquakes.

Private Companies (News Media, Fact-Checking Organizations)

  • Automation of Fact-Checking
    • Conduct manual fact-checking before article creation using this system to reduce the work period.

Examples and Use Cases

Use Case 1. Authenticity Judgment of Information Posted on Social Media

During the Noto Peninsula earthquake, various false information was disseminated on social media, causing confusion and problems. This information contains incorrect content due to carelessness, misunderstanding, or malice. For such information, people who notice the problem can add warning information and its evidence, or automatically collect and add evidence from the internet that the information is incorrect (or correct). Fujitsu's technology enables people who view the post to utilize the added information to assess the credibility of internet information and understand the evidence.

Use Case 2. Support for Fact-Checking Work

Trustable Internet supports fact-checking businesses and companies in verifying and exploring the accuracy of information they collect. The addition of evidence information, whether added automatically or manually, reduces the workload of fact-checking and enables efficient verification. Additionally, by quickly identifying information that suggests misinformation, companies can soon detect misinformation on SNS and enhance their risk management. Fujitsu's anti-fake technology allows the provision of accurate information and a quick response.

Use Case 3. Lead to Fact-Checking Articles

Fact-checking businesses can add information containing links to fact-checking articles to social media posts and news articles, allowing many internet users to access these articles. Fujitsu's anti-fake technology enables fact-checking businesses to provide verification results on the accuracy and validity of information to a broader range of internet users.

Technical Trial

We have prepared a way for you to experience this technology.

Press Releases and News

Tech Blog

Academic Publications

  • Kenji Oki, Kazuhiro Yamashita, and Shinya Kitajima, "A Hybrid Approach Combining LLMs and Web-Based Information for Automated Fact-Checking," in 2025 IEEE International Conference on Web Services (ICWS), July, 2025, pp. 960-962, doi: 10.1109/ICWS67624.2025.00125. (peer-reviewed)
  • Mebae Ushida, Takeshi Miyamae, Takeru Fukuoka, Yoshiyuki Sakamaki, Yoshitomo Sakuma, Kota Miyake, and Takuya Sakamoto, "An Information-Sharing System that Guarantees the Authenticity and Privacy of Contributors," in 27th International Conference Information Security and Cryptology (ICISC), November, 2024, pp. 379–400, https://doi.org/10.1007/978-981-96-5566-3_19. (peer-reviewed)