In the fast-paced world of cybersecurity, the use of open-source large language models (LLMs) has become essential for startups and industry leaders to stay ahead of evolving threats. These models provide a cost-effective and adaptable solution to combat cyberattacks that are constantly evolving.
Recent announcements at the RSAC 2025 conference by companies like Cisco, Meta, and ProjectDiscovery have highlighted the importance of open-source LLMs in shaping the future of cybersecurity. The shift towards collaborative efforts in creating unified defenses against complex threats has been a key takeaway from the event.
Cisco’s Foundation AI group, developed in partnership with Robust Intelligence, is leading the way in creating domain-specific AI infrastructure tailored for cybersecurity applications. Their innovative approach, using a cybersecurity-specific dataset, has resulted in the creation of Foundation-sec-8B, an open-weight Large Language Model with 8 billion parameters.
By open-sourcing Foundation-sec-8B, Cisco is enabling cybersecurity providers to collaborate in strengthening their defenses against cyber threats. The model’s versatility allows for integration into existing tools and the development of new workflows tailored to each organization’s unique needs.
Meta, on the other hand, has expanded its AI Defenders Suite to enhance security in generative AI infrastructure. Their open-source toolkit now includes Llama Guard 4, a multimodal classifier for detecting policy violations across text and images. Additionally, Meta has introduced LlamaFirewall, a real-time security framework with modular capabilities to protect AI agents and detect vulnerabilities in generated code.
ProjectDiscovery, recognized as the “Most Innovative Startup” at the RSAC Innovation Sandbox, offers Nuclei, an open-source vulnerability scanner that rapidly identifies vulnerabilities across various platforms. Their success underscores the value of open-source models in democratizing security and driving innovation in cybersecurity.
Overall, the collaborative efforts of companies like Cisco, Meta, and ProjectDiscovery are reshaping the cybersecurity landscape. By embracing open-source AI tools and specialized domain expertise, cybersecurity providers can enhance their defenses and deliver better security efficacy at a fraction of the cost. The future of cybersecurity lies in openness, collaboration, and innovation, as highlighted by the recent developments at RSAC 2025.