Skip to content

Home

Multi-AI agent security technology to protect against vulnerabilities and new threats
Automating Security Operations with AI
Industry's First Framework to Protect Systems from Threats

Challenges in AI Security

Challenges in AI Security
The surge in cyberattacks and over 20,000 new vulnerabilities annually are exacerbating the shortage of security professionals.
Traditional countermeasures struggle to keep pace with increasingly complex systems and evolving AI-powered attacks.

Fujtisu's Multi-AI agent security technology

Fujtisu's Multi-AI agent security technology
Proactive response to new threats through automated multi-AI agent security technology. A highly versatile solution leveraging knowledge from diverse systems, strengthening the security of AI services like Kozuchi and Takane, and significantly reducing response times.

AI Security Demo Video

Overview of Features

Target Industries and Users

  • CISO organizations (information security control, xSIRT, SOC departments)
  • On-site security operators (system administrators)
  • Users of Fujitsu AI services (Kozuchi, Takane)

Challenges in Target Industries and Users

  • Increasing number of new threats (vulnerabilities +20,000/year, AI-powered attacks/attacks on AI)
  • Attack code development techniques are evolving, requiring immediate patching of vulnerabilities to prevent breaches (IPA Top 10 Threats, 0-day attacks, the increasing exploitation of SSL VPN vulnerabilities is one example of 0-day attacks)
  • Without intervention, security operational workload will become unsustainable.

Technical Issues

  • Limited publicly available information on new threats makes it difficult to comprehensively create the attack and defense scenarios necessary for preventing attacks.
  • While generative AI is rapidly spreading, new attacks (such as prompt injection) are also increasing, necessitating new technologies to address vulnerabilities.

Value of AI security technology

  • Countermeasures deployed within hours of threat discovery, before patch release.
  • Reduces on-site security response costs by 1/3.
  • Differentiates Fujitsu AI services through enhanced security.

Fujitsu's Technological Superiority

  • Industry's first AI security framework enabling proactive defense even by non-experts.
  • LLM vulnerability scanner capable of automatically and comprehensively checking the security resilience of LLMs. Supports over 3,500 vulnerabilities, industry-leading.
  • The LLM guardrails work in conjunction with the LLM vulnerability scanner to ensure the safe and secure operation of generative AI by automatically applying guard rules that detect and reject malicious prompts deemed to require action during IT system operations, thereby preventing inappropriate responses.

Use Cases

  • End-users
    • On-site system administrators understand the details of new vulnerabilities, investigate their impact on their systems, and take preventative measures to prevent breaches.
  • Application developers
    • Automatically perform security testing of developed LLM applications.

Case Studies

  • We will commence field trials of the technology for generative AI security enhancement in partnership with Cohere Inc., beginning December 2024.

Technical Trial