CyCraft Unveils XecGuard to Defend Against LLM Threats

CyCraft, a leading AI cybersecurity vendor, officially launched XecGuard, a plug-and-play guardrail security module designed to defend against large-scale language models (LLMs). The launch of XecGuard marks a significant turning point toward the realization of safe and trustworthy AI, addressing the serious security challenges associated with the rapid adoption of LLMs.

Why Trustworthy AI is Important

The revolutionary power of large language models (LLMs) also brings with it significant security uncertainties. Security threats to LLMs, such as prompt injection, prompt extraction, and jailbreaking, have been discovered, and companies must protect their AI models from these malicious attacks. Traditionally, AI security measures have been treated as “optional.” This has resulted in serious consequences, such as the AI itself and valuable data being exposed to threats, leading to the leakage of confidential information, reduced service stability, and ultimately a loss of customer trust.

CyCraft emphasizes that “AI security should not be optional but implemented as standard,” and believes it is essential to realize stable, reliable, and advanced services.

Also Read: CTC Launches Darktrace AI for Network Threat Detection

The Need for AI Defenses to Address Imminent Risks

The importance of immediate and effective AI security protection grows every day. As AI becomes more integrated into the core operations of businesses, the external attack surface expands, making active defense essential.

CyCraft leverages its “deep battle-tested expertise” in critical sectors such as government, finance, and high-tech manufacturing to accurately address emerging technology threats. The development of XecGuard marks a shift from an era of “using AI to solve cybersecurity challenges” to an era of “using AI to protect AI,” embodying the future of building security and resilience into security measures from the start.

“AI security should be standard, not optional,” said Benson Wu, co-founder and CEO of CyCraft, emphasizing XecGuard’s high defensive capabilities and integration of defense experience cultivated in critical fields.

“In the past, we used AI to tackle cybersecurity challenges, but now we’re using AI to protect AI,” said co-founder and CTO Jeremy Chiu, adding that XecGuard enables businesses to adopt trustworthy AI and use it to deliver high-quality services.

Co-founder and CISO PK Tsung said, “XecGuard enables companies to build security and resilience into their systems from day one,” highlighting its importance in building the world’s most advanced AI security platform.

XecGuard helps ensure safe AI deployment

CyCraft has globally launched XecGuard, a plug-and-play guardrail security module designed to defend large-scale language models (LLMs). XecGuard provides robust defense against prompt injection, prompt extraction, and jailbreak attacks, enabling the high security and resilience required for enterprise-level AI models.

XecGuard can be instantly integrated into any LLM without architectural changes, is easy to deploy, and provides powerful, autonomous defenses right out of the box. XecGuard can be delivered as a SaaS solution, as an OpenAI-compatible LLM defense module on your cloud (e.g., AWS or Cloudflare Workers AI), or embedded on-premise in your custom NVIDIA-based LLM server.

XecGuard has also undergone rigorous testing on major open source models, including Llama 3B, Qwen3 4B, Gemma3 4B, and DeepSeek, and has consistently demonstrated high security while maintaining core performance. Even small-scale models offer a level of protection comparable to that of large-scale commercial-grade models.

Proven results: Working with NVIDIA partner APMIC, XecGuard was integrated into the F1 open source model, resulting in an average 18.5% improvement in overall security defense scores and up to 55.3% improvement in specific attack scenarios during LLM red team exercises.

By combining XecGuard with LLM services, companies can add enterprise-level AI security to their LLM. This cutting-edge AI technology from CyCraft accelerates the adoption of reliable and resilient AI across all industries, enabling organizations to confidently deploy AI, protect sensitive data, and innovate with confidence.

SOURCE: PRTimes