Press Releases 2025
SoftBank Corp. Unveils “Agent Firewall” Prototype to Enable Secure Collaboration Between AI Agents
Aims to establish a next-generation data platform that balances data sovereignty with security
July 24, 2025
SoftBank Corp.
SoftBank Corp. (“SoftBank”) announced the successful development of a prototype of its “Agent Firewall” (tentative name) solution designed to enable AI agents to exchange data safely and flexibly under differing corporate and organizational data-handling rules and governance standards while preserving data sovereignty and regulatory compliance (patent pending). This technology allows data owners—whether nation-states or corporations—to manage and protect their data according to applicable laws and regulations, even as AI agents share information across corporate and internal boundaries.
Background and objectives
In recent years, the trend toward utilizing multi-AI agent systems—where multiple AI agents collaborate to optimize broad societal and business functions—has accelerated. While communication standards for AI agent protocols, data formats, and interfaces are evolving, cross-organizational data sharing still faces significant challenges. For example, differences in data residency, access-control rigor, and bespoke management policies can all impede secure and compliant collaboration.
To meet these challenges, SoftBank has developed its “Agent Firewall” prototype, which enforces and monitors AI agent communications in strict accordance with each organization’s governance rules.
Key features of “Agent Firewall” (tentative name)
| Dynamic governance control | Automatically adjusts inspection depth based on real-time risk assessments and required security levels. |
|---|---|
| AIAgent authentication & authorization | Permits communication only between verified, trusted AI agents. |
| Contract & policy enforcement | Detects and blocks violations of contractual terms and internal policies, incorporating proactive Data Loss Prevention *1 (DLP). |
| Monitoring, auditing & log management | Records all communications in a tamper-resistant ledger for real-time oversight and post-event auditing. |
SoftBank’s “Agent Firewall” complements existing standards—such as A2A*2 (Agent-to-Agent Protocol) and ACP*2 (Agent Communication Protocol)—ensuring consistent security policy application regardless of protocol type. Moving beyond traditional network perimeter-based security*3, it adopts a zero-trust approach*4: every message is continuously verified, and checks are performed only as needed to maintain security while minimizing system load. This design scales to large, distributed network environments.
Future developments
SoftBank will accelerate development and validation, aiming for practical deployment in the fiscal year 2026 (ending March 31, 2027). Looking further ahead, integration with Sovereign Cloud and Sovereign AI environments will enable enterprises, government bodies, and research institutions to safeguard highly sensitive data—paving the way for the realization of next-generation social infrastructure.
SoftBank remains committed to advancing R&D and driving real-world implementation of cutting-edge technologies to build a secure and trustworthy AI-driven society.
Enabling secure, trusted collaboration among AI agents
- [Notes]
-
- *1Data Loss Prevention (DLP): Systems that prevent the loss or external leakage of confidential or critical data.
- *2A2A & ACP: Protocols enabling interoperability among diverse AI agents.
- *3Network Perimeter Security: Traditional model distinguishing between “inside” and “outside” of a network to prevent unauthorized access.
- *4Zero Trust: Security framework based on continuous verification and evaluation, with no implicit trust granted to any request.
- *1
- SoftBank, the SoftBank name and logo are registered trademarks or trademarks of SoftBank Group Corp. in Japan and other countries.
- Other company, product and service names in this press release are registered trademarks or trademarks of the respective companies.