---
title: supa.ai begins same-day penetration testing for AI model cyber attacks
url: https://www.dataloco.com/en/supaai-begins-same-day-penetration-testing-for-ai-model-cyber-attacks
published: 2026-10-14T16:21:24+00:00
language: en
section: Security
source: https://ascii.jp/elem/000/004/441/4441834/?rss
organizations: supa.ai
publisher: Dataloco
---

# supa.ai begins same-day penetration testing for AI model cyber attacks

supa, inc. has launched same-day penetration testing for cyber attack simulations using its proprietary AI models. This service, known as "Security Check AI," enables the execution of tests on systems using attack methods that exploit Fable/Astra level frontier AI and open-weight models. The tests are conducted securely on supa's GPU servers, ensuring that sensitive information and detected vulnerabilities are not transmitted to external services or cloud servers.

The cost for these penetration tests ranges from 500,000 yen to 1,200,000 yen (excluding tax), substantially lower than traditional manual testing, which can be up to seven times more expensive. Testing scenarios are crafted with the assistance of white hat hackers who consider real-world operations and user permissions, followed by AI conducting the validations while security engineers verify the results and their implications.

Penetration testing attempts to breach systems from the perspective of an attacker, assessing how far critical information and functions can be accessed. supa utilizes a unique infrastructure that combines large language models (LLM) with a harness to control AI execution. Reports from the tests are provided within three business days after the testing scope is confirmed, and the process is designed to ensure confidentiality by not sending diagnostic content to external AI services.

With increasing reports of AI being used in cyber attacks, supa aims to help organizations verify their authentication and permission management against potential breaches. The company’s approach combines expert judgment with AI execution to reduce the financial burden on companies seeking penetration testing services.

supa's infrastructure is designed to operate on local GPU servers, adhering to ISMS standards for governance. This ensures that all information and execution logs related to the systems tested remain secure and confidential. The service supports not only initial scans but also re-evaluations after necessary corrections have been made to address identified vulnerabilities.

As AI technology becomes more integrated into cyber attack strategies, supa’s new service aims to provide timely and effective solutions for companies looking to safeguard their digital environments. The company’s mission focuses on reducing Japan's digital deficit by creating affordable, AI-native software solutions and facilitating a collaborative environment for engineers to work with AI.

## This story in other languages

- [العربية](https://www.dataloco.com/ar/shrk-supa-ttlk-khdm-fhs-amny-fory-yaatmd-aal-althkaaa-alastnaaay-lmhaka-alhgmat-alsybrany)