
RSA Conference 2025: Key Themes and Highlights
From the rise of AI-powered threats to the complex regulatory landscape, every major theme at RSAC this year pointed to a critical need: organizations must have visibility and control over who and what has access to what — and when.

In this article
AI dominated RSA Conference 2025, not just as a security tool, but as a new attack surface. Defenders and attackers alike are harnessing AI’s power. From the rise of AI-powered threats to the complex regulatory landscape, every major theme at RSAC this year pointed to a critical need: organizations must have visibility and control over who and what has access to what — and when.
Whether human users or non-human identities (like AI agents), ensuring the right access at the right time has never been more vital. That’s where autonomous identity comes in.
If you missed out on this exciting event, don’t worry – we’ve got you covered. We’ve put together a full recap of the three-day event below, featuring the key themes and highlights you need to know.

Introducing Lumos Autonomous Identity
First off, RSAC 2025 marked a significant milestone for Lumos, as a de facto coming out party for Lumos Autonomous Identity. The conference emphasized the crucial role of identity security in a world where AI and digital transformations are constantly evolving, so it was a natural fit to announce our new platform and capabilities.
We announced the launch of our platform earlier in the week and were able to have many engaging conversations about autonomous identity at our RSA booth.

Our Founder and CEO, Andrej Safundszic also gave a theatre presentation on autonomous identity and how it can move your organization beyond the constraints and frustrations of legacy IGA.

Andrej covered what makes Lumos autonomous identity different from legacy IGA, including how it:
- Improves Time-to-Value
- Stops Access Sprawl
- Ends Rubber-Stamping
- Cuts the Cost of Ownership
Book a demo now to learn more!
Key Themes From RSAC 2025
RSAC is always a major event in security, and the 2025 edition was no different. Thousands of security practitioners and professionals met in San Francisco to exchange perspectives, strategies, and insights on the future of security and the unique challenges the industry faces in 2025.
Some of the key themes from the conference included:
- The Dual Role of AI in Cybersecurity
- Identity remains the (new) perimeter
- Emergence of Agentic AI
- Regulatory Landscape and Privacy Concerns
- Collaboration and Community Building
1. The Dual Role of AI in Cybersecurity
One of the most discussed and debated themes at RSA Conference 2025 was the double-edged nature of artificial intelligence in cybersecurity. In the right hands AI is a tool while in the wrong, it is a weapon.
While AI-powered real-time detection and predictive capabilities offer new advantages, attackers now use AI to automate phishing, exploit overprovisioned access, and impersonate users with deepfakes. The takeaway? Identity is the new perimeter. Managing access with precision is no longer optional — it’s critical for stopping AI-fueled attacks before they start.
AI as a Defensive Tool
On the defense front, AI is finally living up to the promise the industry’s been making for the last decade. We’re seeing real, tangible applications of AI that go beyond buzzwords—particularly in real-time threat detection, vulnerability discovery, and malware analysis.
Vendors and researchers showcased how AI models are being trained on vast telemetry data to identify anomalous behavior across endpoints and cloud environments faster than human analysts ever could. This translates into early detection of zero-day attacks and rapid triage of alerts that used to overwhelm SOC teams.
Additionally, AI is proving invaluable in proactive security. Several tools now use machine learning to predict potential vulnerabilities based on code patterns, usage behavior, or even geopolitical context. Automated malware analysis has also taken a leap forward—with AI dissecting and classifying threats at machine speed, reducing manual effort and response time.
AI as a Threat Vector
But what’s helping defenders can just as easily help attackers. And that’s the catch.
RSA 2025 didn’t shy away from spotlighting how threat actors—particularly advanced persistent threat (APT) groups and cybercriminal syndicates—are now weaponizing AI to scale their operations. We’re talking AI-enhanced malware that adapts on the fly, deepfake-driven social engineering, and phishing campaigns generated and personalized using generative AI.
Speakers warned that AI doesn’t just make attacks faster; it makes them smarter. The line between real and fake is blurring, and traditional security awareness training might not be enough in the face of ultra-targeted, AI-generated deception.
Bottom line: AI is reshaping the cybersecurity landscape. Whether that shift favors defenders or attackers will depend on how rapidly security teams can evolve. The arms race is officially underway.
2. Identity Is the New Perimeter—For Real This Time
One theme that stood out at RSA 2025 was that identity is the new perimeter—and it’s not just a trendy soundbite. It’s a strategic imperative. The traditional security perimeter has long since dissolved under the weight of SaaS sprawl, hybrid work, and increasingly sophisticated attackers who don’t break in, they log in.
Discussions throughout the three days made it clear: defending the perimeter starts with defending identity. Credential theft, privilege escalation, and lateral movement were recurring topics, and for good reason. Once an attacker is inside with stolen or weak credentials, the damage multiplies fast.
Strong identity security is now foundational. That includes passwordless login, MFA, behavioral analytics, and strict identity mapping across systems.
To reduce risk, companies are prioritizing:
- Enterprise password managers
- Unique, complex passwords
- Regular audits of privileged accounts
- Elimination of stale credentials
But stopping at strong passwords isn’t enough. The next layer is stopping attackers from escalating privileges or moving laterally once they’re inside. That’s where Zero Trust architectures, least-privilege access, and Privileged Access Management (PAM) come into play.
Bottom line: In 2025, protecting your organization means treating identity like your first—and most important—security boundary. It’s not just “back in style.” It’s the only way forward.
3. Emergence of Agentic AI
At RSA Conference 2025, one of the more thought-provoking topics was the rise of agentic AI—autonomous systems capable of making independent decisions without human intervention. This isn’t the stuff of sci-fi panels or academic debate anymore. It’s here, it’s operational, and it’s raising serious questions for security leaders.
Agentic AI represents a shift from traditional automation and rule-based systems to self-directed models that can adapt to new environments, evaluate outcomes, and initiate actions on their own. Think of an AI that not only identifies a threat but decides how to contain it, isolates a system, reroutes traffic, and notifies stakeholders—all without a human touching the console.
Naturally, this capability offers massive upside in environments where speed and scale are essential. But it also opens the door to new forms of operational and ethical risk. At RSA, conversations turned quickly to the “what ifs”: What if an agentic system takes the wrong action? Who’s accountable? What guardrails need to be in place to prevent unintended consequences—or exploitation by adversaries?
Several speakers emphasized the urgent need for governance frameworks to oversee the deployment of these technologies. Transparency, auditability, and human-in-the-loop policies were flagged as critical requirements—not optional features. As agentic AI continues to evolve, security teams will need to walk the line between autonomy and control with far more precision than ever before.
As agentic AI takes hold, your identity governance must evolve from gatekeeping to continuous, autonomous control. Static roles and outdated access review processes won’t cut it when AI agents can request and act on access autonomously.
Bottom line: agentic AI isn’t just another tool in the belt—it’s a new kind of operator. And it needs to be treated that way.
4. Regulatory Landscape and Privacy Concerns
At RSA Conference 2025, it became clear that keeping pace with evolving global regulations is no longer just a compliance checkbox—it’s a full-time strategic priority. As governments and regulatory bodies around the world continue tightening their grip on data protection, organizations are feeling the squeeze. And not just in legal departments—this is squarely on the radar for CISOs, security architects, and DevOps teams alike.
The conference underscored the growing complexity of the regulatory landscape, with frameworks like GDPR, CCPA, Brazil’s LGPD, and newer entrants from APAC and Africa shaping how data must be handled, stored, and protected. With enforcement actions on the rise and fines climbing, there’s now a direct link between regulatory hygiene and reputational risk.
New global regulations make proving who has access — and why — essential. From privacy mandates to audit readiness, identity governance with access reviews and visibility are front and center. Organizations need modern governance with AI-powered explanations, audit-ready records, and automatic policy enforcement - delivering clarity in natural language.
Intersection of privacy and security
Privacy-by-design emerged as more than a buzzword—it’s becoming a foundational design principle for modern cybersecurity programs. Speakers advocated for embedding privacy considerations early in development cycles, rather than tacking them on as an afterthought. This includes everything from minimizing data collection and using anonymization techniques to ensure data localization compliance and strong access controls.
One clear takeaway: privacy and security are no longer separate disciplines. The most forward-thinking organizations are unifying them into a single strategy, with governance frameworks that are agile enough to adapt as laws evolve. Compliance isn’t static—and your systems can’t be either.
Bottom line: Whether you're building a new app or auditing legacy infrastructure, RSA 2025’s message was loud and clear: regulatory expectations are rising, and the margin for error is shrinking. Security leaders need to stay informed, stay adaptable, and most importantly, build systems that respect user data by design—not just by policy.
5. Collaboration and Community Building
RSA Conference 2025 wasn’t just a showcase of bleeding-edge tech and bold predictions—it was a reminder that cybersecurity is, at its core, a team sport. This year’s theme, “Many Voices. One Community.”, hit home for many attendees, highlighting a message that’s increasingly hard to ignore: we’re only as strong as our collective defense.
Identity is at the core of this collaboration. From cross-industry threat sharing to integrated JML (joiner-mover-leaver) workflows that reduce friction across teams, autonomous identity helps organizations move faster and smarter — together.
Stronger and faster together
Threat actors don’t operate in silos—so defenders can’t afford to either. Keynote speakers urged the community to move beyond the outdated model of proprietary knowledge hoarding.They encouraged radical transparency, where sharing indicators of compromise (IOCs), attack patterns, and even response playbooks became the norm—not the exception. Public-private partnerships, intelligence sharing across industries, and open communication between governments, vendors, and practitioners were all flagged as critical in the fight against increasingly sophisticated cyber threats.
There was also a strong push to make the cybersecurity industry more inclusive and diverse, both in terms of background and thought. The challenges ahead—AI misuse, quantum risk, regulatory upheaval—aren’t just technical. They’re human, social, and global. And solving them will require a broader spectrum of voices, skill sets, and lived experiences.
Bottom line: no vendor, no government agency, and no company is going to solve this alone. Community isn’t a nice-to-have—it’s our best defense.
Closing Thoughts on RSAC 2025
RSA 2025 spotlighted the challenges ahead — AI threats, quantum risks, regulatory demands. But it also revealed a clear path forward: a world of agentic AI, autonomous identity security and the importance of community and collaboration in cybersecurity.
AI is redefining what’s possible on both sides of the security equation. Regulations are evolving faster than compliance teams can blink. And attackers are working together more effectively than ever. The message for security leaders and defenders is clear: adaptability, visibility, and collaboration remain critical for survival.
This year’s theme, “Many Voices. One Community.”, wasn’t just conference branding—it was a call to action. If we want to build resilient, ethical, and future-proof systems, we have to start by building a stronger, more connected security community. Whether you're deploying AI agents, embedding privacy into design, or just trying to stay ahead of the next threat vector, you don’t have to go it alone.
And at Lumos, we’re proud to be part of that future. RSA 2025 marked the debut of Lumos Autonomous Identity, and the response was incredible. With real-time visibility, dynamic policy enforcement, and AI-driven insights, Lumos is eliminating the manual processes and outdated legacy IGA systems that have held teams back for too long.
Ready to see how autonomous identity can transform your approach? Book a demo today!










