Meeting AI-driven attacks with AI-driven defense: Teleport is approved for access to Anthropic’s most advanced AI model, Claude Mythos 5.1

Passed the review for the Cyber Verification Program, for which publicly announced cases in Japan remain limited. Teleport will put it to work in research and development for Data is CITY, which already runs in-house, and is recruiting about five initial partners for CITY for Enterprise, an offering for companies

Teleport Inc. (CEO: Tomoyasu Hirano), a startup that develops and puts AI agents to work, has passed the review for the Cyber Verification Program (CVP) offered by Anthropic, the U.S. AI developer, and has been approved for access to Claude Mythos 5.1, Anthropic’s most advanced AI model.

Claude Mythos 5.1 is an AI model with exceptional capabilities in finding and analyzing software vulnerabilities (security weaknesses). Because those capabilities could become a weapon if used for attacks, Anthropic has held off on a general release and offers the model only to users who have passed reviews such as the Cyber Verification Program.

Anthropic has announced that, using an early version of Mythos (Claude Mythos Preview), it found thousands of high-severity vulnerabilities, including ones in every major operating system and web browser. In Japan, Hitachi, Ltd. has announced that its evaluation through Project Glasswing confirmed that, for source code running to millions of lines, Mythos (Claude Mythos Preview) could build threat models in a matter of hours, work that previously took security specialists and developers weeks (July 2026).

So far, only a small number of companies in Japan have announced that they have been approved for access.

Teleport will use Claude Mythos 5.1 on its own systems and in research and development for Data is CITY, which already runs in-house. Building on the knowledge and technology gained there, Teleport will bring companies of every size a way to understand their digital environment together with AI agents and keep improving it.

Bringing the advanced analysis capabilities of Claude Mythos 5.1 to Data is CITY

With this approval, Teleport can now use the advanced reasoning and analysis capabilities of Claude Mythos 5.1 on its own systems and in research and development for Data is CITY.

By combining the structure of the entire digital environment that Data is CITY captures with Mythos’s in-depth analysis of individual software and systems, Teleport will identify compound security risks that were previously hard to find, and make its countermeasures more sophisticated.

Finding a vulnerability does not, on its own, prevent damage. Countermeasures move forward only once it is clear which of the company’s systems it touches, which operations it affects, and who will fix what, in what order. Through Data is CITY, which depicts an entire company as a single “city,” and through developing AI agents that work in dialogue with people, Teleport is building the system that takes care of what comes “after it is found.”

Teleport is also developing technology in which multiple AI agents work together to provide continuous support, from investigating vulnerabilities to planning countermeasures and verifying the results of fixes.

Through this research and development, Teleport will strengthen the technology foundation of Data is CITY. It will offer custom system builds to companies with complex IT environments, and an ongoing subscription service that keeps the burden of adoption and operation low to mid-sized and small businesses.

CITY for Enterprise: Teleport is recruiting about five initial partners

Teleport is recruiting about five initial partner companies for CITY for Enterprise, its offering for companies and organizations.

For each company, Teleport’s AI agents work with the company’s staff to turn the digital environment of the entire group into a CITY. Within the first three months, it is shaped into something the executive meeting can look at, and after that, improvement and the building of know-how continue together. Because Teleport goes deep with one company at a time, the number of partners is limited.

Teleport also welcomes inquiries about CITY for Business, its offering for mid-sized and small businesses.

On request, Teleport also holds workshops and talks for executives and board members on security in the age of AI.

Data is CITY: people and AI agents running the same city

Teleport runs its own digital environment on CITY. Developed in-house by Teleport, CITY is a technology that organizes a company’s digital infrastructure across vendor and system boundaries and visualizes it as a single “city.”

It manages the relationships among servers, networks, databases, accounts, access permissions, and more in a common data structure, and represents them as a 3D city.

In that city, AI agents developed in-house by Teleport continuously investigate and keep track of the digital environment, understanding how the systems are structured, what role each plays, and how they relate to one another.

Executives and employees can talk with the AI agents directly.

“What does this server do?”
“If this system goes down, which operations are affected?”
“Where should we start improving?”

The AI agents explain based on actual system information and propose improvements. People understand what they hear, make the decisions, and carry out the improvements together with the agents.

Data is CITY is not just a tool for visualizing IT environments. It is a system in which people and AI agents understand the same world and keep updating the digital environment through dialogue.

Beyond security, it supports improvements across a company’s entire digital environment, including IT asset management, understanding operating costs, and organizing systems.

The PHP security issue: response plan clear in about five minutes, fix completed within 30 minutes

When the PHP security issue emerged in October 2026, Teleport, through dialogue with its AI agents, understood the impact on its own systems and the response needed in about five minutes, and completed its countermeasures within 30 minutes (according to company records).

This response was possible because Teleport already had the full picture of its digital environment and had shared that information with its AI agents before the issue arose.

Rather than starting to investigate after a problem occurs, Data is CITY lets you understand your environment day to day and keep improving it. That is one of its defining features.

Background: attacks get in through the “connections”

According to Japan’s National Police Agency, 226 ransomware incidents were reported in 2025, and by size, about 60% of the affected companies and organizations were small and medium-sized businesses. In a survey of affected organizations, more than half said recovery cost them JPY 10 million or more in total.

In the same survey, VPN devices accounted for more than 60% of intrusion routes. Attackers get in by exploiting unpatched vulnerabilities, leaked credentials and weak passwords, misconfigurations, and the like, then explore the internal network until they reach important data and backups. In an attack on a major beverage maker in September 2025, the attackers entered a data center via network equipment within the group, and product ordering and shipping were halted (National Police Agency, “Threats in Cyberspace in 2025” (in Japanese, March 2026)).

Each weakness may be small on its own, but following the connections between systems can lead to the core of the business. And as advanced AI spreads to attackers as well, the time it takes for a weakness to be found and exploited is likely to keep getting shorter.

Yet in many companies, websites, servers, cloud services, business systems, and more are each managed by different departments and vendors, and no one has an integrated view of how they connect and depend on one another. When a problem occurs, it takes time to investigate the scope of impact and decide on countermeasures. Teleport is developing a system that treats a fragmented digital environment as a single shared world that people and AI agents can understand and improve together.

Comment from Tomoyasu Hirano, CEO

“The further digitalization goes, the less executives can grasp the full picture of their own company’s systems. I don’t think that is how it should be.

“In a factory or a store, executives go to the floor, talk with the people who work there, understand the problems, and make improvements. But the digital space alone is left entirely to specialists and vendors, and executives themselves no longer know what is going on. Before it is a security problem, this is a management problem.

“What Teleport wants to change with Data is CITY is this structure itself. People and AI agents understanding the same ground, talking with each other, and improving it day by day. We will turn the digital environment into a place where executives and employees can take an active part.

“This approval is a major step toward making that happen. We want to take the world’s most advanced knowledge and technology that we gain here to companies and local communities all over Japan, whatever their size.”

Comment from Tsubasa Kawakami, CTO and Security Lead

“With the arrival of AI, attackers can now probe companies’ systems at scale in a short time. A small misconfiguration, a permissions gap, or outdated software may each be minor on its own, but combined, they become a way in.

“Data is CITY draws the full picture of how a company’s sites, servers, accounts, and permissions are connected. Claude Mythos 5.1 has the ability to dig deep into a single piece of software or an app within that picture, and to verify whether a vulnerability can actually be exploited. Connections across the whole, and the depth to dig into a single point. By combining the two, we aim to find the paths from small weaknesses to important data before attackers do.

“From investigation to planning a response and checking afterward, multiple AI agents work together and do it again and again. We will make it possible to carry out advanced security work, which used to take specialists a great deal of time, faster and on a continuous basis.

“We will make this approval an opportunity to take the technology foundation of Data is CITY a big step forward.”

About the Cyber Verification Program

The Cyber Verification Program (CVP) is a review-based program through which Anthropic provides access to advanced AI capabilities to security professionals and organizations that meet its requirements.

The program was expanded on October 6, 2026, and now has three access categories based on use, covering areas such as defensive vulnerability analysis, malware analysis, and incident response.

Official information: https://www.anthropic.com/news/cyber-verification-program

Inquiries

Applications for the CITY for Enterprise partner program and inquiries about CITY for Business are accepted here: https://teleport.jp/dataiscity/en/#contact

About Teleport Inc.

Teleport Inc. is a Japanese AI startup whose work includes developing an AI agent platform, generative AI education, and helping companies and local communities put AI to use.

Its products include Telepotch, AI agents with a name and a face that work in LINE and Slack groups and remember each person. Under the motto “Turning everyday life into an adventure,” Teleport aims to build a society where people and AI agents act together.

Website: https://teleport.jp/en/
Data is CITY: https://teleport.jp/dataiscity/en/