#Japan #Tokyo #AI Data #Deepfake #SYNTHETIQ VISION

Combatting DeepFake Crimes with Innovative AI Technology

In recent times, the misuse of DeepFake technology has surged globally, presenting new challenges. This manipulation often results in fake videos impersonating celebrities or worse, creating content where children's faces are digitally altered to produce pornographic material. Such actions lead to severe issues such as defamation and breaches of privacy. Moreover, incidents of online fraud involving eKYC (Electronic Know Your Customer) processes, stock market manipulation, and misinformation campaigns have become prevalent, shaking the foundational belief that video content reflects objective reality. Thus, there is an urgent necessity for a reliable system to objectively assess the authenticity of video content.

AI Data Corporation's Initiative

In response to these challenges, AI Data Corporation, headquartered in Tokyo and led by CEO Takahito Sasaki, is utilizing the DeepFake detection AI technology known as "SYNTHETIQ VISION," developed by the National Institute of Informatics (NII). The company has established an internal verification environment to assess the effectiveness and operational methods of its AI video authentication platform.

Currently, AI Data Corporation is in the "verification phase" as it evaluates various aspects, including:

• - Assessment trends on sample videos assuming criminal scenarios and issues.
• - A comparison of these findings with traditional visual and manual verification processes.
• - Consideration of legal evidential strength and the integration potential into internal controls and audit processes.
• - Planning operational flow in collaboration with investigation agencies, municipalities, and businesses.

The timeline and form of commercial service release will be determined based on these verification outcomes.

Understanding SYNTHETIQ VISION

"SYNTHETIQ VISION" is an advanced AI developed by the NII, specifically designed to identify fake videos with the following technical characteristics:

• - It analyzes unique "subtle generation traces" left on synthetic facial videos, outputting a score of authenticity.
• - It employs a numerical evaluation method for videos that are difficult for the human eye to discern.
• - The model aims to maintain detection performance even after video compression or re-encoding.
• - It features an architecture that stores judgment results and related logs, enabling retrials and references during auditing processes.

AI Data Corporation diligently investigates how to make "video authentication independent of human judgment" feasible, how to integrate this system with existing visual checks and fraud detection systems, and how to design management and retention rules for judgment logs that encompass evidence and accountability aspects. Note: This press release does not guarantee the performance of "SYNTHETIQ VISION"; it simply announces that verification using this technology has commenced.

Projected Use Cases and Social Impact

The AI video authentication platform is anticipated to be leveraged across several domains, contingent on future verification results and discussions with relevant organizations:

• - Law Enforcement/Education Boards:

- Assisting in initial screenings of potentially defamatory videos or pornographic content created without consent from images of children and students.
- Aiding in the organization and documentation of evidential videos during victim reports.

• - Municipalities/Government Agencies:

- Supporting the verification of videos related to impersonation or misinformation disseminated through consultation hotlines.
- Preparing technical documentation for deletion requests to platform operators.

• - Financial Institutions/eKYC Providers:

- Mechanically identifying potentially suspicious videos involving impersonation or synthetic faces during online identity verification processes.
- Scoring for high-risk transactions or additional validation flows.

• - Media Companies/Social Media Platforms:

- Employing AI scores in pre-distribution checks and secondary evaluations upon reporting.
- Visualizing risks related to brand defamation and misinformation spread.

These scenarios represent potential applications, and operational strategies will be individually determined after thorough discussions and validations with experts in each sector.

Future Plans

AI Data Corporation aims to enhance and validate the AI video authentication platform through several steps:
1. Testing with Near-Practical Cases: Evaluating data reflecting real-world scenarios and identifying challenges.
2. Collaborative Verification with Stakeholders: Engaging in joint trials with law enforcement, educational institutions, banking sectors, and local governments.
3. Establishing Operational Guidelines: Ensuring consistency with legal evidential standards and creating compliance regulations.
4. Integrating Services: Exploring methods to connect with existing data recovery and investigation services.

AI Data Corporation is committed to minimizing the damage caused by the misuse of DeepFake technology by striving for a practical and empirically validated AI video authentication platform. In the long term, they envision the system for technical validation of video authenticity as a vital part of social infrastructure, fostering collaborations with relevant authorities and research institutions.

About AI Data Corporation

• - Name: AI Data Corporation
• - Established: April 2015
• - Capital: 100 million yen (Capital reserve 1.525 billion yen)
• - CEO: Takahito Sasaki
• - Location: 4F, Metro City Kamiyacho Building, 5-1-5 Toranomon, Minato-ku, Tokyo
• - URL: www.aidata.co.jp

AI Data Corporation has been dedicated to protecting and utilizing data assets of companies and individuals for over 20 years, backed by a strong reputation from over 10,000 businesses and more than a million customers. With a robust "data ecosystem business" that encompasses data sharing, backup, recovery, migration, and deletion, the company has achieved the top-selling rank for 16 consecutive years at the BCN Awards.

In addition to cloud data management and recovery services, AI Data is esteemed in the legal domain for providing forensic investigations and evidence disclosure services, which have received recognition from the Minister of Economy, Trade, and Industry. Moreover, through initiatives like the development of the patent search system 'Tokkyo.Ai' and a marketplace for buying and selling patents, the company supports intellectual property management and monetization.