Cloud, Assurance, Forensics, Engineering

Tag: agenticAI

AI is here to stay? One person’s perspective on attending Datatribe – Cyber Innovation Day 2025

AI Security Takes Center Stage: Key Insights from DataTribe’s Cyber Innovation Day 2025 by Shawn Anderson, CTO and 2x CISO, Boston Meridian Partners

November 4th’s industry gathering revealed how artificial intelligence is fundamentally reshaping cybersecurity – from autonomous red teams to agentic AI governance

I was already a fan of DataTribe, but their daylong event at The Capital Turnaround—a historic Navy Yard car barn turned vibrant event venue—solidified my admiration. With engaging speakers, impressive startups, dynamic attendees, and great food, the event was a standout. Located in a revitalized area near the Washington Navy Yard, this venue is a must-see for your next event.

DataTribe’s Cyber Innovation Day 2025 brought together cybersecurity’s brightest minds to tackle the most pressing question facing our industry: How do we secure systems that are increasingly powered by artificial intelligence? From startup pitches to expert panels, the day revealed both unprecedented opportunities and sobering challenges ahead.

The AI Revolution in Security: Faster, Smarter, More Dangerous

The opening presentations from DataTribe’s portfolio finalists painted a picture of AI’s transformative impact. Anit Saeb, founder of Cytadel and former head of penetration testing at the Bank of England, demonstrated how AI-driven autonomous red teaming can achieve “full compromise in under 8 minutes—550x faster than ransomware groups.” According to Cytadel’s internal testing, his company’s AI agents have already bypassed the top three EDR vendors, proving that traditional defenses are struggling to keep pace.

Meanwhile, Tim Schultz from Starseer (formerly Verizon’s AI Red Team lead) highlighted a critical gap: “Current AI security tools only monitor user-LLM interactions, while agents act across databases and applications and communicate with other agents.” As organizations deploy AI agents that can independently access systems and make decisions, we’re entering uncharted territory for security governance.

The scale of this challenge became clear through Evercoast’s presentation on physical AI training. Their platform addresses a fundamental problem: “Physical AI has only thousands of hours of training data vs trillions for LLMs.” As AI systems move from chatbots to controlling physical infrastructure—from F-16 repairs to autonomous vehicles—the security implications multiply exponentially.

Industry Veterans Sound the Alarm

Jason Clinton, Deputy CISO at Anthropic, provided a sobering insider perspective on AI’s current trajectory. “AI compute [is] increasing 4x year-over-year since 1957,” he noted, with Anthropic now writing “~90% of code via Claude.” But this acceleration comes with risks: “Threat actor capability compression [is] occurring—Tier 1 and Tier 2 actors are converging as script kiddies can now ask models to write ransomware and C2 infrastructure.”

The shift in workflow is fundamental. As Clinton described it, we’re moving to “ask AI to do work, return to check results”—a complete reversal of traditional development processes. This creates new categories of vulnerabilities that traditional security tools weren’t designed to handle.

Dmitri Alperovitch, co-founder of CrowdStrike, brought historical perspective to these challenges. Reflecting on CrowdStrike’s founding after the 2010 Operation Aurora attacks, he emphasized that “if you can stop sophisticated actors, everything else becomes easy.” His advice for today’s founders was characteristically direct: “Don’t fear big company competition – fear unknown hungry startups.”

The Investment Landscape: Opportunity Amid Uncertainty

The investment panel featuring Rob Ackerman, Andrew McClure, and Phil Venables revealed a market in transition. “2025 cybersecurity financing: ~1,000 events, $15B volume with 50% being AI/AI-first companies,” they reported, but warned that “Series A to B graduation [is] declining (400 A rounds vs 40 B rounds = 10:1 ratio).”

The key insight? We’re moving from “orchestration” to what they termed “choreography” – AI agents organizing themselves in ways that traditional human-managed systems never could. This shift requires entirely new approaches to security architecture and governance.

Security Leaders Grapple with the “Lethal Trifecta”

Security practitioners Maurice Boissiere, Randy Sabett, and Pat Moynahan introduced a crucial framework for AI security risk assessment. They identified the “Lethal Trifecta for AI Agents: external data sources, external communications, and private data via unprompted input.” This framework provides a practical lens for evaluating AI deployments, though they admitted the overall assessment remains “chaotic due to AI adoption pressure vs security fundamentals.”

The panel emphasized that while “C-suite [is] now paying attention,” many organizations still lack basic incident response capabilities, with insufficient logging and “no forensic capabilities to determine breach scope.”

Media and Market Reality Check

Daniel Whitenack from the Practical AI Podcast provided valuable context on AI’s evolution, identifying three distinct phases: traditional ML (still widely used for specific tasks), foundation models (requiring technical expertise), and current generative AI that’s “squeezing out the middle” by enabling “business domain experts [to] bypass data scientists.”

Maria Varmazis from T-Minus Space Daily highlighted sector-specific vulnerabilities, noting that the “$614B global space industry” remains “10-15 years behind cybersecurity best practices.” Recent incidents include University of Maryland researchers using an “$800 antenna to intercept sensitive military/police communications” and “Russia’s 2022 ViaSat attack [that] disabled Eastern European satellite communications.”

Startup Innovation: Hardware Meets AI

Beyond software solutions, Tensor Machines demonstrated how AI security extends to physical systems. With “$2M NSF funding and 5 patents filed,” they’re addressing the “$5T+ autonomous systems market” through “physics-informed neural networks for real-time physical fingerprinting.” Their live demonstration showed automatic failover when camera spoofing was detected – exactly the kind of autonomous response needed as AI systems become more prevalent in critical infrastructure.

Lessons from the Trenches: Fundraising and Building

Throughout the day, practical wisdom emerged from battle-tested entrepreneurs. Alperovitch’s fundraising philosophy resonated: “Would you rather have 50% of a pea or 10% of a watermelon? No one ever went bankrupt because of dilution.” His emphasis on execution over technology trends – “customers buy effectiveness, not technology trends” – provided grounding amid AI hype.

The bourbon tasting session offered its own metaphor for startup persistence, featuring Charleston Red Corn Bourbon made from a “colonial-era variety that nearly died out” until a “Clemson professor found 2 cobs in seed vault [and] regenerated the line.” Sometimes the most valuable innovations come from reviving what others have given up on.

Take Action: Preparing for the AI Security Future

The insights from DataTribe’s Innovation Day point to several immediate actions every cybersecurity leader should take:

Assess your AI exposure now. Use the “Lethal Trifecta” framework to evaluate every AI deployment in your organization. Catalog which systems have external data access, communication capabilities, and access to private data without human oversight.

Invest in behavioral detection over signatures. Traditional signature-based security is already failing against AI-generated threats. Companies like Tensor Machines are pioneering behavioral fingerprinting approaches that can adapt to new attack patterns in real-time.

Prepare for agent governance. Whether you’re deploying AI agents or defending against them, establish clear policies for agent identity management, permission structures, and audit trails. The companies that solve this challenge early will have significant competitive advantages.

Bridge the talent gap strategically. With AI democratizing both offensive and defensive capabilities, focus on hiring people who can architect secure AI systems rather than just operate traditional security tools. The future belongs to organizations that can “choreograph” rather than just orchestrate their security operations.

Plan for autonomous security. As Jason Clinton noted, we’re approaching a world where “AI writes code → AI finds bugs → AI tests vulnerabilities → AI fixes issues.” Start experimenting with AI-powered security automation in low-risk environments to build competency for this inevitable future.

The cybersecurity industry stands at an inflection point. Organizations that act on these insights now—while their competitors are still debating whether AI is hype or reality—will be the ones defining security standards for the next decade. The question isn’t whether AI will transform cybersecurity, but whether you’ll be leading or following that transformation.

I attended DataTribe’s Cyber Innovation Day 2025 and compiled insights from presentations, panels, and networking sessions throughout the event.

Please reach out to us via our webpage and LinkedIn below.

www.bostonmeridian.com

Boston Meridan LinkedIn Page <- Follow this company!

About the author

Shawn Anderson has an extensive background in cybersecurity, beginning his career while serving in the US Marine Corps. He played a significant role as one of the original agents in the cybercrime unit of the Naval Criminal Investigative Service.

Throughout his career, Mr. Anderson has held various positions, including Security Analyst, Systems Engineer, Director of Security, Security Advisor, and twice as a Chief Information Security Officer (CISO). His CISO roles involved leading security initiatives for a large defense contractor’s intelligence business and an energy company specializing in transporting environmentally friendly materials.

Beyond his professional achievements, he is also recognized for his expertise in the field of cybersecurity. He is a sought-after speaker, writer, and industry expert, providing valuable insights to both C-Suite executives and boards of directors.

Currently, Mr. Anderson serves as the Chief Technology Officer (CTO) for Boston Meridian Partners. In this role, he evaluates emerging technologies, collaborates with major security providers to devise cybersecurity strategies, and delivers technological insights to the private equity and venture capital community.

Overall, Shawn Anderson’s career journey showcases a wealth of experience in cybersecurity and leadership roles, making him a respected and influential figure in the industry.

Observations so far in 2025 – Data, AI, and everything new under the sun.

Every spring and late summer, the Boston Meridian Partners team attends two of the world’s largest cybersecurity conferences: RSA Conference in San Francisco and Blackhat in Las Vegas. These events are a whirlwind of activity—client receptions, dinners, and nearly 250 meetings with innovative tech companies and industry leaders. This year, I spent much of my time in Blackhat’s “Startup City,” a hub of emerging companies that’s far easier to navigate than the sprawling Moscone Center.

The Startup Scene: Shiny Pennies and Hidden Gems

Startups at these conferences are eager to showcase their innovations, often presenting themselves as the “next big thing.” However, many struggle to stand out in a crowded market. While their enthusiasm is infectious, differentiation is key—what I call the “shiny penny problem.” A penny, no matter how polished, is still a penny if it doesn’t offer unique value.

Meeting C-level executives at startup booths was a highlight, as this is rare for larger companies where leaders like George Kurtz or Satya Nadella are often mobbed by media. These interactions offered valuable insights into emerging technologies, particularly in artificial intelligence (AI). AI and Data: The Heart of Modern Security

AI dominated conversations this year, with startups focusing on data-driven security solutions. At Blackhat’s AI Summit, I heard repeated emphasis on the importance of data and identity in building secure environments. As I’ve said for decades, “It’s all about the data.” The rise of Large Language Models (LLMs) has amplified this, with data being consumed at unprecedented rates. However, the lack of controls—such as a Cloud Access Security Broker (CASB) for LLMs—raises concerns about rogue models masquerading as legitimate tools.

CISO Challenges: Balancing Priorities in a High-Pressure Role

Discussions with Chief Information Security Officers (CISOs) at Blackhat and RSA Conference revealed the immense pressure they face in balancing day-to-day operations with the need to adopt cutting-edge technologies. Many CISOs described their roles as a constant exercise in “blocking and tackling”—managing fundamental security tasks like patching vulnerabilities, responding to incidents, and ensuring compliance. These operational demands often leave little time to explore emerging technologies like AI-driven security tools or advanced identity management platforms.

One CISO from a manufacturing company shared a striking perspective: they prioritized keeping the production floor operational over implementing a robust data protection strategy. “If the factory stops, the business stops,” they explained, noting that downtime could cost millions. While understandable, this approach undervalues the long-term risks of data breaches, where sensitive intellectual property or customer data could be compromised. For example, a single unprotected dataset could be exfiltrated by attackers, leading to regulatory fines or reputational damage far exceeding the cost of a temporary production halt.

Identity management emerged as another significant challenge. Several CISOs reported using two to four different identity solutions, creating complexity and potential security gaps. One CISO from a financial services firm described their struggle with integrating legacy systems with modern cloud-based identity platforms, resulting in fragmented visibility into user access. They expressed frustration with “platform” solutions that promise seamlessness but often fall short, leading them to favor “best-of-breed” tools. However, this approach can increase costs and administrative overhead. A better strategy, as I’ve advocated previously, is adopting one or two well-integrated identity solutions that work seamlessly across on-premises and cloud environments, reducing complexity while maintaining robust security.

The growing prevalence of Internet of Things (IoT) and Operational Technology (OT) devices is also keeping CISOs up at night. With networks hosting tens of thousands of devices—ranging from smart sensors in offices to industrial control systems in factories—securing these endpoints is a daunting task. A CISO from a utility company highlighted the challenge of monitoring “dumb” devices with outdated firmware alongside “smart” IoT devices that are often poorly configured. They noted a recent incident where an unpatched IoT camera served as an entry point for a ransomware attack, underscoring the need for solutions that can monitor and secure both on-premises and cloud-connected devices.

Finally, many CISOs admitted to feeling overwhelmed by the rapid pace of technological change, particularly in AI. One CISO from a healthcare organization confessed they lacked the bandwidth to evaluate AI-driven security tools, relying instead on their team’s recommendations. This highlights a broader issue: CISOs are expected to be strategic visionaries while managing tactical firefights, often without sufficient resources or time to stay ahead of the curve.

Emerging Trends in Cybersecurity: Why Staying Current Matters

The cybersecurity landscape is evolving rapidly, driven by advancements in AI, the proliferation of connected devices, and increasingly sophisticated threats. Staying current with these trends is critical for CISOs and their teams to protect their organizations effectively. Falling behind can lead to blind spots, such as unaddressed vulnerabilities or missed opportunities to leverage new tools for efficiency and resilience. Based on my observations at Blackhat and RSA Conference, here are seven key technological trends shaping the industry, along with why staying informed is essential:

  1. AI Agent Governance: As organizations deploy AI agents for tasks like threat detection and customer support, the lack of oversight frameworks creates new risks. Rogue or misconfigured AI agents could expose sensitive data or disrupt operations. For example, a poorly governed AI chatbot might inadvertently leak proprietary information. CISOs must adopt “AI Agent Rewind” capabilities to audit and recover from misuse. Staying current ensures organizations implement governance early, avoiding costly mistakes as AI adoption scales.
  2. Detection-as-Code Revolution: Traditional manual security rule creation is giving way to AI-powered platforms that auto-generate detection rules from threat intelligence. Companies like SOC Prime now serve over 1 billion rules globally, enabling faster responses to emerging threats. CISOs who fail to adopt these tools risk falling behind adversaries who exploit automation. Keeping up with this trend allows organizations to scale their security operations efficiently, especially in resource-constrained environments.
  3. External Attack Surface Expansion: Threats increasingly bypass traditional perimeter defenses, requiring active validation of vulnerabilities “outside the firewall.” Tools that simulate real-world attacks are replacing passive scanning, providing more accurate risk assessments. For instance, a retailer recently discovered an exposed API through active testing, preventing a potential breach. Staying informed about this trend helps CISOs prioritize external risks, which are often overlooked in favor of internal network security.
  4. Behavioral Security Over Signatures: Signature-based detection is losing ground to behavioral fingerprinting and drift analysis, which establish custom baselines for each application to achieve near-zero false positives. A bank using behavioral security detected an insider threat by identifying unusual data access patterns, avoiding a significant breach. CISOs who embrace this trend can reduce alert fatigue and focus on real threats, but staying current is critical to selecting the right tools for their unique environments.
  5. LLM Firewall Emergence: As enterprises integrate Large Language Models (LLMs) into workflows, new tools are emerging to protect against prompt injection, data leakage, and model manipulation. For example, a healthcare provider recently faced a prompt injection attack that tricked an LLM into revealing patient data. LLM firewalls can mitigate these risks, but CISOs must stay educated on this nascent category to implement effective controls before widespread adoption.
  6. Browser-Based Threat Protection: Zero-day phishing attacks that bypass email security are a growing concern. Browser-based tools using computer vision and real-time analysis are protecting over 800,000 users by detecting malicious sites instantly. Staying current on this trend allows CISOs to bolster endpoint security, especially for remote workforces increasingly targeted by sophisticated phishing campaigns.
  7. Unified Security Platforms: Talent shortages and budget constraints are driving demand for platforms that consolidate IT, InfoSec, and cybersecurity functions. A unified platform enabled a mid-sized firm to reduce its security tools from 15 to 3, cutting costs and improving visibility. CISOs who stay informed about consolidation trends can streamline operations and address the “do more with less” mandate, but failing to keep up risks reliance on outdated, fragmented solutions.

Staying current with these trends is not just about adopting new tools but about understanding how they align with organizational priorities and constraints. For CISOs juggling operational demands, dedicating time to research, attending conferences, or collaborating with peers is essential to avoid being blindsided by new threats or missing opportunities to enhance security posture. Organizations that invest in continuous learning—through training, industry reports, or vendor partnerships—will be better equipped to navigate the complexities of modern cybersecurity.

Looking Ahead

As AI, data, and IoT/OT reshape the threat landscape, CISOs must balance innovation with foundational security practices. Staying ahead requires not only technical expertise but also strategic foresight to prioritize what matters most. I’d love to hear your thoughts—what trends are you seeing at conferences, and how are you keeping up? Share your insights below or connect with us at

www.bostonmeridian.com

Boston Meridan LinkedIn Page <- Follow this company!

About the author

Shawn Anderson has an extensive background in cybersecurity, beginning his career while serving in the US Marine Corps. He played a significant role as one of the original agents in the cybercrime unit of the Naval Criminal Investigative Service.

Throughout his career, Mr. Anderson has held various positions, including Security Analyst, Systems Engineer, Director of Security, Security Advisor, and twice as a Chief Information Security Officer (CISO). His CISO roles involved leading security initiatives for a large defense contractor’s intelligence business and an energy company specializing in transporting environmentally friendly materials.

Beyond his professional achievements, he is also recognized for his expertise in the field of cybersecurity. He is a sought-after speaker, writer, and industry expert, providing valuable insights to both C-Suite executives and boards of directors.

Currently, Mr. Anderson serves as the Chief Technology Officer (CTO) for Boston Meridian Partners. In this role, he evaluates emerging technologies, collaborates with major security providers to devise cybersecurity strategies, and delivers technological insights to the private equity and venture capital community.

Overall, Shawn Anderson’s career journey showcases a wealth of experience in cybersecurity and leadership roles, making him a respected and influential figure in the industry.