TrendAI™, the enterprise AI security brand of Trend Micro (TSE: 4704), a global leader in security, announced on June 3 that it has officially joined the Project Glasswing project led by Anthropic. This collaboration will use Anthropic’s yet-to-be-publicly released cutting-edge model Claude Mythos Preview to strengthen software code review and vulnerability analysis capabilities and help threat intelligence researchers transform AI-driven vulnerability discovery results into actual risk reduction measures.
This is not a pure technical cooperation. When Anthropic officially launched Project Glasswing on April 7 this year, it assembled technology giants such as AWS, Apple, Google, Microsoft, NVIDIA, CrowdStrike, Palo Alto Networks, etc., with only one goal: to use the vulnerability discovery capabilities of AI models to conduct defensive scans for critical software around the world before these capabilities spread into the hands of malicious actors. The joining of Trend Micro adds another heavyweight member to the alliance with nearly 40 years of practical experience in information security.
Project Glasswing: Why is Anthropic gathering technology giants for information security?
Project Glasswing It was born out of a disturbing discovery: the Claude Mythos Preview model trained by Anthropic has surpassed most human hackers in its ability to understand program code and discover vulnerabilities. This is not an exaggeration. According to data released by Anthropic, Mythos Preview has discovered thousands of high-severity vulnerabilities in critical software including every major operating system and browser. One such case was the discovery of a 27-year-old vulnerability in OpenBSD, which is known for its security.
The British AI Security Institute (AISI) conducted an independent assessment of Claude Mythos Preview’s network security capabilities on April 13, confirming that its vulnerability discovery capabilities have indeed reached an unprecedented level. Anthropic bluntly stated in an official statement: “AI models have reached a level of programming ability that can surpass all but the best humans in discovering and exploiting software vulnerabilities.”
The problem is that this ability won’t always be in the hands of law-abiding parties. As AI technology continues to advance, similar vulnerability discovery capabilities will soon proliferate. Anthropic therefore decided to take the initiative and prioritize the use of Mythos Preview capabilities to organizations responsible for critical infrastructure, allowing them to scan and patch first. Only after the vulnerabilities have been properly patched can the use of the model be gradually expanded.
To this end, Anthropic has committed up to US$100 million (approximately NT$3 billion) in Mythos Preview usage, and donated an additional US$4 million (approximately NT$120 million) to open source security organizations. Project Glasswing is currently available to more than 40 organizations responsible for building or maintaining critical software infrastructure.
Claude Mythos Preview: Thousands of high-risk vulnerabilities can be discovered in “a very short time”
Claude Mythos Preview is a cutting-edge model created by Anthropic specifically for information security scenarios. It has not yet been officially released to the public and is only available to specific organizations through Project Glasswing. Commenting on the project, Minister of Digital Development Lin Yijing pointed out that Claude Mythos “discovered thousands of high-risk (critical) security vulnerabilities in a very short period of time, and its vulnerability discovery capabilities have even exceeded the scale that the traditional hacker community can discover in a year.”
More importantly, this model can not only find a single vulnerability, but can also connect multiple vulnerabilities in series to simulate potential attack paths and attack chains. This means that it can discover not just individual weaknesses, but the complete attack surface, which is far more valuable to defenders than traditional vulnerability scanning tools.
Anthropic stated in the Glasswing progress update announced on May 22 that partners have discovered specific CVE vulnerabilities (such as CVE-2026-5194) through Mythos Preview, confirming the system’s vulnerability discovery capabilities in the real world.
TrendAI’s role: From threat intelligence to collaborative vulnerability disclosure
TrendAI™’s role in Project Glasswing is more than just a “user”. As the enterprise AI security brand of Trend Micro, TrendAI™ will use Claude Mythos Preview to strengthen its capabilities in three major areas:
- Software code review and analysis: Use Mythos’ code understanding capabilities to accelerate the security assessment of critical software
- Collaborative vulnerability disclosure and prioritization: Convert the vulnerability results discovered by AI into executable patching plans to help software vendors prioritize the highest-risk issues
- Vulnerability protection and virtual patching: Provide real-time protection through the Trend Vision One platform to achieve quantifiable risk reduction before official patches are released
Rachel Jin, chief operating officer of TrendAI™ Enterprise Business Group at Trend Micro, said: “We are closely aligned with Anthropic’s goal of using AI to make all software more secure. Project Glasswing is an important opportunity for us to explore how advanced AI technology can help software vendors find vulnerabilities earlier and further improve the security and resilience of the systems that customers rely on every day.”
Taiwan perspective: The next step in AI hardware revitalization is AI software security
This cooperation is of special significance to Taiwan. Chen Yihua, CEO and co-founder of Trend Micro, pointed out that Taiwan has a complete and world-leading AI supply chain. Many important international companies and AI industry customers work closely with Taiwan’s industry chain, so they need the most advanced AI technology and security protection capabilities. She emphasized that as Taiwan is a global center for AI hardware, the next stage of growth momentum will come from AI software and application innovation. Trend Micro will use the most advanced AI security technology to support more Taiwanese companies and start-ups to accelerate AI application deployment and operational models.
Trend Micro has also recently launched the “Inception Program” to use the most advanced AI technology and security capabilities to assist the development of Taiwan’s AI startups. It hopes to jointly enhance the security and competitiveness of Taiwan’s AI industry through the world’s leading Frontier AI and AI Security technologies.
Minister of Digital Development Lin Yijing started from the perspective of national security. He said that Taiwan is the center of global AI hardware. If critical infrastructure or AI supply chains suffer major security attacks, it will affect the security of the global technology industry and democratic countries. “Information security protection is an important issue for Taiwan and the world.” He is happy to see Trend Micro successfully join Project Glasswing and hopes that Trend Micro will help Taiwanese companies discover and patch vulnerabilities early and strengthen Taiwan’s overall AI and security defense capabilities.
The Defender’s Advantage: Anthropic’s head of security talks about the core of collaboration
Rob Bair, head of cyber security cooperation at Anthropic, made a key judgment when talking about this cooperation: “In the past 30 years, the core of information security has been the offensive and defensive competition of human intelligence. Now artificial intelligence has entered both offense and defense. Attackers are using AI to find vulnerabilities faster and create more convincing scams. Therefore, the key is no longer whether AI will shape the future of information security, but who can use AI most effectively, and the answer is defenders.”
Rob Bair emphasized that Anthropic brings one of the most advanced AI models in the world, but the model itself does not mean defense, nor does it understand the enterprise’s network environment. “This is exactly the value that Trend Micro brings: the deep security expertise accumulated on the front line for decades, and the trust relationship established with customers around the world.”
He specifically mentioned that Trend Micro has experienced every major transformation in the information security industry in the past 40 years, from the earliest computer viruses to the Internet, the cloud, and now AI, and has always been a leader in the global information security field. When the AI era comes, Trend Micro chose to become one of the first companies to transform into an AI-native company, integrating AI into research, operations, and defense methods, “using AI to fight malicious AI.”
By leveraging Anthropic’s models more deeply, Trend Micro will combine its leading threat intelligence and telemetry data accumulated from protecting hundreds of thousands of organizations and millions of users to create stronger defense capabilities. Rob Bair said: “The future of information security will be written by those who have the courage to protect it. Trend Micro has such courage, and Anthropic is very proud to walk side by side with you.”
Conclusion
The emergence of Project Glasswing marks the official upgrade of AI’s role in the security field from “auxiliary tool” to “core combat capability.” When AI models can discover thousands of high-risk vulnerabilities in a very short time, even surpassing the results of a human hacker community in a year, the rules of the game for information security defense have been completely rewritten.
The joining of Trend Micro brings nearly 40 years of practical experience and global customer trust to this alliance. For Taiwan, the ability of local information security companies to stand on this international stage also confirms Taiwan’s strategic value in the field of AI security. As AI capabilities continue to evolve rapidly, whether defenders can stay ahead of attackers will determine the safety of the entire digital world. Project Glasswing is where this race begins.
Source: KOCPC Chinese