Boundless Launches Defensive Access: Dedicated AI Models for Authorized Teams to Combat Surge in Attacks

Source: Boundless

AI inference service provider Boundless today announced the launch of Boundless Defensive Access, a program that opens up a specially adjusted Abliterated (low rejection) model for vetted security teams. The first model to go live is a security research version of the open-source model DeepSeek V4.1 Flash 1017. This program aims to address a long-standing challenge faced by defenders: general AI models often reject legitimate security testing requests that are similar in form to attack vectors, while attackers are not subject to such constraints.

Defenders Should Not Fall Behind Attackers in Tools

The security industry is facing an unprecedented volume of attacks. Penetration testing, vulnerability validation, and attack simulation inherently require defenders to think from the attackers' perspective. However, these requests are often highly similar in form to real attacks, and even when security teams are testing systems they own and are authorized to test, general models frequently refuse to assist.

This has led to malicious actors being able to deploy and modify open-source models on their own, while compliant defense teams find themselves at a disadvantage in terms of tools. The goal of Defensive Access is to provide defenders with equivalent AI capabilities under strict access controls.

First Model: DeepSeek V4.1 Flash 1017

Independent evaluation agency Vals AI released early benchmark results on October 20, showing that the standard version of DeepSeek V4.1 Flash 1017 outperformed GPT-6 Sol in its multi-step cybersecurity tasks based on simulated enterprise networks. Building on this, Boundless offers a security research version of the model that reduces unnecessary rejections in legitimate security testing scenarios.

To meet the high-precision real-time security task demands, Boundless provides top-tier inference performance support:

• Over 300+ Tokens/second: Providing rapid responses for security automation agents and real-time penetration testing tools.
• Top throughput: Measured speeds of up to 349.90 tokens/sec on the Vercel AI Gateway, ranking first in the industry.
• Enterprise-grade hosting: Eliminating the heavy burden of deploying and maintaining large models for security teams, ensuring 24/7 stable operation.

"The security restrictions of traditional models often do more harm than good, hindering those who genuinely want to provide protection," said the Boundless team. "With Defensive Access, we are equipping defenders with unrestricted cutting-edge model capabilities through a high-throughput inference engine."

Four Core Application Scenarios and Application Mechanism

Boundless Defensive Access focuses on four defensive scenarios:

• Code Review: Identifying and validating potential security vulnerabilities before system deployment.
• Defense Testing: Simulating real hacker techniques to identify blind spots in existing defenses.
• Penetration Testing: Conducting in-depth probing and validating remediation plans within authorized scopes.
• AI Application Testing: Testing the defensive capabilities of AI applications against malicious inputs and jailbreak commands.

To ensure the technology is used safely and responsibly, Defensive Access adopts a strict application review system. Each applying team must undergo compliance qualification checks to ensure the model is used only for compliant security research, red team exercises, and penetration testing. Additionally, for enterprises with special needs, the Boundless team also supports customized deployments of other open-source Abliterated models.

How to Apply

Boundless Defensive Access is now officially open for applications. Security practitioners and enterprise teams can visit the official website to submit their applications.

About Boundless

Boundless is a high-performance AI inference infrastructure platform focused on open-source models, dedicated to providing ultra-low latency and high-throughput model hosting and inference services for AI-native enterprises and developer teams.

This article is contributed and does not represent the views of BlockBeats

This content is provided for general informational purposes only and doesn't constitute financial, investment, legal, or tax advice. Any events, rewards, online promotions, or related information mentioned herein should not be considered a recommendation, solicitation, or invitation to purchase, sell, trade, or otherwise deal in any crypto assets. Crypto assets are highly volatile and may result in loss. The availability of WEEX services, products, and related events may vary by region. You are responsible for ensuring that your participation is in accordance with applicable local laws and regulations.