The cybersecurity landscape is undergoing a major shift as tech giants like Cisco and Meta champion open-source artificial intelligence (AI) solutions. At RSAC 2025, these companies showcased how new large language models (LLMs) and toolkits can be used for advanced cyber defense. Their efforts signal a turning point, and now we can say that open-source AI is no longer a fringe experiment but a core pillar of modern cybersecurity infrastructure.
RSAC 2025: The Stage for Open Innovation
The RSA Conference 2025 (RSAC), held in San Francisco from 28th April to 1st May, marked a pivotal moment for the cybersecurity industry. With over 41,000 attendees, 700 speakers, and 450+ sessions, this year’s event spotlighted the accelerating convergence of artificial intelligence (AI), digital identity, and collaborative defense.
Cisco’s Foundation-sec-8B: LLM for Cyber Defense
Cisco made headlines with the release of Foundation-sec-8B, an open-source large language model specifically trained for cybersecurity applications. It has leveraged Meta’s Llama 3.1 architecture and, unlike general-purpose LLMs, Foundation-sec-8B is tailored to understand and respond to security-specific queries, analyze threat intelligence, and automate incident response tasks. With this new launch, the company intends to make AI-driven security accessible to organizations of all sizes, enabling faster detection and remediation of cyber threats.
According to the company, Foundation-sec-8B has been tested on different in-house built datasets, such as vulnerability databases, threat intelligence reports, security tooling documentation, threat behavior mappings, etc. Further, the company claims that the LLM can be tuned to different threat intel, telemetry, and detection rules.
According to Jeetu Patel, Chief Product Officer, Cisco, “Foundation-sec-8b is a foundational step toward building AI-native security systems tools that don’t just process data but truly understand the security domain. This model gives security teams everywhere a powerful building block to accelerate defense, reduce fatigue, and gain clarity in complex threat environments.”
Meta’s AI Defenders Suite: Modern Threats Guard
Alongside Cisco, Meta also announced the extension of the AI Defender Suite by adding Llama Guard 4 and LlamaFirewall. Apart from that, it also introduced:
- PromptGuard 2: for detection of prompt injections and jailbreak attempts
- Agent Alignment Checks and CodeShield: for protection against AI agent decision-making processes and related vulnerabilities
- CyberSec Eval 4: developed in partnership with CrowdStrike, which is an open-source benchmarking suite
ProjectDiscovery and Nuclei: Open-Source Vulnerability Scanning
The “Most Innovative Startup” award at RSAC 2025 was bagged by ProjectDiscovery for its open-source vulnerability scanner Nuclei. The innovative vulnerability scanner is capable of scanning threats across websites, networks, APIs, and cloud environments.
Moreover, with over 3,000 contributors and a growing template repository, Nuclei shows how open-source tools can be scaled to meet real-world security demands for both SMBs and Fortune 500 companies.
What’s Next In The Line Of Innovation
The Foundation-sec-8B by Cisco is just the beginning of the new era. The company has announced that under its Cisco Foundation AI program, it will keep launching more such cybersecurity innovations. Some of the releases that it has already announced are a cybersecurity reasoning model and a benchmark suite for the evaluation of AI models on real-world tasks.
It is also expected that going forward, the market players will focus on enhancing model robustness, expanding community governance, and integrating AI more deeply into hardware and network layers.