Video: Post-Predict 2025 Recap | Top 10 Takeaways for Intelligence-Driven, Autonomous Defense | Duration: 3612s | Summary: Post-Predict 2025 Recap | Top 10 Takeaways for Intelligence-Driven, Autonomous Defense | Chapters: Intelligence to Action (100.525s), Autonomous Threat Operations (214.465s), Top 10 Takeaways (316.62s), Emulating Adversary Tactics (363.27s), Continuous Third-Party Monitoring (506.91998s), AI Empowering Analysts (595.92s), Enriching Security Alerts (830.86005s), Collaborative Threat Intelligence (980.14996s), Cybercriminal PR Tactics (1282.7899s), Business Value of Intelligence (1424.9401s), Continuous Threat Monitoring (1574.61s), Q&A and Recap (1872.555s)
Transcript for "Post-Predict 2025 Recap | Top 10 Takeaways for Intelligence-Driven, Autonomous Defense":
Welcome to the Intelligence to Action webinar. I'm Kathleen Kuzma. And I'm Maddie Malitz. We are coming to you live from the Recorded Future headquarters to talk about cybersecurity, or as CISOs call it, another reason to sleep with Microsoft Teams open. Predict twenty twenty five just wrapped in New York City and London where thousands of intelligence leaders gathered to discuss how to turn intelligence into action. Because here's the thing, a lot of companies are dealing with serious data overload right now. They're getting data from everywhere, from threat feeds, security tools, vendor alerts, logs. And while they're trying to sort through all of it to figure out what actually matters, cyberattacks are only becoming more frequent and advanced. Check out what Recorded Future CEO Colin Mahoney said in his opening remarks. Everyone I talk to in this space has the same overall vision. You want the intelligence to work as fast as the adversaries. You wanna turn that insight into action instantly. You wanna prove the value of every intelligence investment you're making. But the attackers have already automated. They're not held back. They're not constrained. And, of course, they're using AI. Of course, they're using large language models. They're moving at machine speed. So you've mastered that art of seeing threats. Now it's time to master the art of stopping them automatically. Imagine if we could focus on what we as human beings do best, namely think, strategize, predict. At the same time, let's let the machines do what they do best, hunt, detect, and actually prevent. Colin is, of course, referring to autonomous threat operations, which recorded feature unveiled app predict. In a nutshell, Autonomous Threat Operations includes enhanced threat hunting, prevention, and detection capabilities, all powered by custom source ingestion and correlation in the intelligence graph. Recorded Future's chief product officer, Jamie Sajak, explains it here. Autonomous threat operations is not a separate platform. It's actually an intelligence orchestration layer that sits on top of your existing recorded future deployment. At the core is the intelligence graph, continuously updated with over 1,000,000,000 events every day. Autonomous threat operations doesn't just flag these 1,000,000,000 events. It roots intelligence to action in your existing controls like firewalls and your SIEM, making your data actually work for you. The future of intelligence is now. And as Colin Mahoney put it, intelligence alone is not enough. The future of intelligence is not just about seeing more threats. There's always gonna be more threats. It's actually it's about stopping them automatically every single time. Really makes you think of all of the other things in life we'd like to stop automatically every single time. The daily voice mail saying you are prequalified for a loan for an oddly specific amount of $59,000, Smishing scammers telling me I'm overdue for my toll payments, and I don't even have an E ZPass account or a car. Those texts from some random number that just says, hey. I mean, the list is truly endless. It really is. But in all seriousness, our industry is constantly battling scammers and cyber criminals with their latest tactics, techniques, and procedures. And that's why we were so thrilled to host so many insightful leaders at Predict to hear how they're fighting back. We listened to all of the sessions, pulled out the biggest insights, and pulled it all down to the top 10 takeaways from Predict 2025. We'll share those insights with you now, and we'll leave some time for your questions at the end. For those questions we don't have time to answer, we can follow-up with you after the webinar. So without further ado, here are the top 10 takeaways from Predict 2025. Takeaway number one. Proactive defense requires knowing your adversary. Analysts from Global Payments and Mastercard showed how their teams emulate attackers to test defenses, From profiling threat actors and simulating their behaviors to guide their defensive strategy, no VR headsets required. And one thing everyone is seeing right now is typosquats and brand image misuse. Here's how Jack Watson's team at Global Payments is fighting back. When it comes to typosquats and brand image misuse, right, a lot of times these platforms and tools look at them together collectively. Right? So you miss things that exist outside of a domain that's typosquat or not typosquat, rather in this case, right, that, still could have, you know, images of your brand on them. Right? And so, you can actually identify some of these, right, by using your similarly scored images, right, like taking examples of where these images exist through the playbook alerting rate and kinda seeing where they live in other spaces. But even the best toolset needs to be stress tested. Banks have the Fed security stress test. Security teams, they have adversary emulation. And Steve Range and Patrick Davey at Mastercard showed exactly how this works in their program. Patrick stressed the importance of bridging the gap between cyber threat intelligence and an actual improvement in your security posture. The team creates and conducts tests using Recorded Future's index group research to make sure that their security controls work because, as Steve puts it, it's a little easier to play defense if you've seen the offenses plays before. Watch this. The big thing is how does Recorded Future help us? Right? You've you've seen the presentations. You've seen some of the stuff that's coming. I'd say the biggest thing for us is the data, the TTP catalogs. Like, we're we're identifying what threat actor we think we need to worry about. We're going out and saying, what's the TTP catalog for these guys? And we're running it through our emulation tools. We identify a threat actor. We research that threat actor in their TTPs. Recorded future is a godsend, in that. With more comprehensive threat intelligence, analysts are now able to run full threat actor simulations at work. The point is, understanding your adversary helps you anticipate their next move and play better defense. Over to you, Maddie. Takeaway number two. Third party risk management needs to be a living, breathing intelligence workflow. Gone are the days of annual vendor reviews. You know the whole check-in twice a year and hope nothing changes in between strategy. But as Mastercard's Kelly White and Recorded Futures Jerry Hodge made clear, third party risk now changes daily between software vulnerabilities, cloud misconfigurations, and geopolitical events. By the time your next quarterly review rolls around, everything is already different. According to Verizon Business's 2025 data breach investigation, 30% of breaches were linked to third party involvement, twice as much as last year and driven in part by vulnerability exploitation and business interruptions. Hodge explained that in order to surface immediate threats across their supply chain, organizations need to take an intelligence driven approach to third party risk management, which includes end to end vendor assessment workflows, continuous and actionable vendor monitoring, predictive insights tuned into your risk context, and automated remediation action and recommendations. You know the old saying trust but verify? This is that verify part. Automated, continuous, and built into your daily operations. With 30% of breaches now involving third parties, your vendor security posture is your security posture, which means verification can't be a twice a year checkbox. It has to be always on. Takeaway number three. AI isn't replacing analysts. It's empowering them. Another key topic at Predict was the evolution of threat intelligence from manual triage to AI assisted decision making. AI co pilots, improved reporting, and natural language querying were all on full display. But the consistent refrain was this, humans still need to lead. Think of AI as the world's most caffeinated research assistant. It can process 10,000 alerts before lunch, but someone still needs to decide which alert's going to warrant a 2AM wake up call. Robert Moody from The Home Depot said that humans can't match the speed of AI enabled attacks, so the only way to scale and get in front of AI is to augment ourselves with AI. He further explained that vulnerability management and prioritization workflows are right for AI automation. But given AI trust issues, the ideal arrangement is for humans to define remediation strategies and scope and then have AI help with the risk scoring and compilation. Robert says, when something anomalous pops up, that's where a person comes in and is really useful in re steering the ship. That's right. Even the smartest AI still needs a captain. Robert reiterated the symbiotic relationship between AI and humans. Take a look. This is the way. It's using AI to break the cycle, to get through this in a sense that you are leveraging AI in a responsible way, not trusting it fully, understanding where it's best put within the intelligence life cycle. Related to Moody's comments, Grammarly's Igor Tarpon and Eric Harbawi described their security organizations, quote, fundamental shift to AI driven tools. Harvawi said, by moving beyond static store playbooks and rigid API calls, we developed a system capable of adaptive reasoning. The real breakthrough came when we integrated Recorded Future, not as a lookup service, but as a dynamic intelligence partner. Igor provided some tactical ways that the team built Recorded Future into their MCP workflows. Take a look. These are example of the query that we use. Some of them are used to start, getting the context into your context window, but some of them are just tactical steps that could be used when you analyze the ticket. And if you're wondering what that looks like, Eric provided an example of a concrete workflow using Grammarly's own MCP with Recorded Future playing an important role. Here is where the most time savings occurs. Recorded features, new AI features are running in parallel. Now instead of taking thirty, sixty minutes on a ticket that might be false positive, what if we reduce that to three minutes? Both Igor and Eric said their new system now has the context to build attribution intelligently and hunt proactively. And they said that in all cases, humans are still in the loop. So here's the takeaway. AI isn't replacing analysts. It's replacing the grunt work, the midnight false positives, the urgent alerts that aren't, the tedious correlation that keeps you from actually threat hunting. Because it still takes a human to lead, ask the right questions, and connect the dots that a machine can't see. Takeaway number four. The noise your SOC ignores may be your next breach warning. Sanjay Kumar of Landis and Gear highlighted a critical truth. Blocked domains, phishing attempts, and low severity alerts are often dismissed, yet they can frequently be traced back to coordinated campaigns. The problem is when you're looking at each alert in isolation, you miss the pattern. Five unrelated low severity events suddenly become one coordinated attack. Sanjay went on to say that enrichment and pattern recognition allows teams to see the big picture that smaller seemingly isolated events create. Take a look. In order to detect those actions, which can be hidden in the background noise, Our job is to sharpen the visibility. There are two steps to sharpen the visibility. The first is enrich the security telemetry. The raw alerts alone do not tell any story. But when we add the context to them, we can turn individual signals into meaningful intelligence. Of course, in a world where vulnerabilities outnumber the hours in your day, not everything can be patched, and perhaps not everything should be. That was a clear message from Recorded Futures' John Bach and doctor Jared Smith who challenged security teams to stop chasing CVSS scores and start focusing on what's actually being exploited in the wild. John says, Recorded Futures attack surface intelligence simulates an attacker's reconnaissance to continuously discover and tie assets to your organization, and it pairs that information with ten plus years of structured datasets to create a complete profile of your attack surface. Then it assesses your assets like a threat actor would, providing real time exposure scoring and AI enriched remediation guidance to help you prioritize patching and save time. The moral of this story, not every vulnerability deserves your immediate attention. Some are serious threats, some look like noise because they are noise, and some look like noise but are actually early indicators of something bigger. That's why enrichment and context are critical. Without them, you're flying blind, unable to distinguish between background noise and the opening moves of a coordinated attack. Takeaway number five. Cross team coordination is critical. In cybersecurity, collaboration often means three teams discovering they've been chasing the same incident. Both the insights impact panel and the keynote presentation by Mastercard's vice president of fusion and intelligence, Michelle McClure, deliver the same message. Intelligence that doesn't flow into operational workflows is useless. If your intel loses context between the SOC and the boardroom, you're not sharing intelligence. You're playing corporate telephone. The best organizations build bridges between threat intel and stakeholders across their entire ecosystem, And Mastercard is taking this even further. Here's what McClure had to say about intelligence driving risk decisions. AI, quantum, and autonomous systems are rapidly changing the landscape. They're reshaping risk. Threat casting is a methodology that we use at Mastercard within the fusion center to forecast threats ten years into the future and beyond. When Mastercard recognized that accelerating global threats demanded a unified intelligence driven response, the company launched its nexus program to connect lines of business, industry partners, threat assessment vendors, governments, customers, and merchants. NexSys leverages real time intelligence to break down communication silos, anticipate emerging risks, and directly link threat intelligence to business outcomes. So what can we learn? Teamwork doesn't just make the dream work. It builds more secure businesses in the short and long run. Michelle's presentation emphasized the criticality of using intelligence to forecast threats well into the future, underscoring the importance of minimizing risk and highlighting the business contributions of cybersecurity, which go well beyond identifying threats. It's all using intelligence for action. This is why the release of autonomous threat operations is so timely. Let's pause for a minute and learn a little bit more about the new product. At Predict this year, our assistant VP of product, Adrian Portescu, walked through a demonstration of autonomous threat hunting capabilities. Check it out. I can see that I have not only intelligence coming from Recorded Future, but it's also coming from FS Isaac, some other third parties, and I also do have indicators that are coming from my own incident response. Recorded future aggregates and enriches all of these custom sources with intelligence and threat context coming from the intelligence graph, making sure that my threat hunts are always up to date at the moment of the execution. As you'll see, the system has automatically brought in all of the behavior rules that are of course suited for threat hunting and related with the threat actors, the malwares, or the TTPs that I am focusing my threat hunt on. Our in situ panelists are not only producing these hunting packages, but also translating and validating the sigma rules so they are ready for deployment at any point. I can actually visualize the original logic of the sigma rule, and I can visualize and validate all of the logic of the translated rules so I know exactly what I'm hunting for. We do translations for SentinelOne SQL, Microsoft KQL, CrowdStrike FQL, and Google Yara l. Adrian continues by showing how quickly analysts can analyze findings from their threat hunts, which can be set up dynamically based on their organization's personalized threat map. Meaning that every hunt includes the latest threats specific to your company. Now let's talk about the three main actions that I can take on my findings. For the high severity ones and the criticals like these two that I have for my CFO's machine, I want to immediately escalate them. So I'll just go ahead and copy the details of these findings so I can put them in the ticket. In the future we'll be able to open up tickets directly from the platform. Next, let's talk about prevention. After I've analyzed all of these findings, I realized that I wanna put some of them in prevention. Luckily, I have connected some of my firewalls to record the future so I can immediately create blocking rules. This powerful compilation of new capabilities is available to our customers with autonomous threat operations. If you'd like to learn more, scan the QR code at the end of the webinar. And with that, let's jump back into the second half of predict highlights. Takeaway number six. Cybercriminals' PR skills can be as dangerous as their hacking skills. In a breakout session, Recorded Futures' Megan Keeling made it clear that threat actors are now actively managing their external brand reputation and expanding their reach using strategic PR tactics. Yes. You heard that right. Cybercriminals are doing media relations now. From direct engagement with media and journalists to sharing exaggerated claims, their goal is to increase their operational effectiveness, influence ransom negotiations, and overcome doubts about whether they'll actually decrypt data or honor agreements. A cybercriminal's reputation can be the difference between a ransom payment in the thousands versus a ransom payment in the millions, Megan said. And this is why recognizing these tactics is essential to building resilience against these attempts to manipulate and influence you. So what can we learn here? If ransomware gangs are running full blown PR campaigns, you can't afford to wing your crisis comms. Have a plan, practice it, and radical idea, actually use it because drafting your first holding statement while you're already trending on the dark web is not the position you wanna be in. Takeaway number seven. Collection without context is just more data. It's true. I went on stage with my colleague, Chris Holden, to talk about Recorded Futures proprietary intelligence collection capability. Let me emphasize this again. Collection, it's only half the equation. Context matters. What makes intelligence actionable is how it's enriched, prioritized, and connected to an organization's mission. This can include identifying victims communicating with active command and control servers via our proprietary network intelligence to alerting on leaked API and AWS tokens on GitHub. So the big lesson here is that having all the data isn't useful if it's not driving a cybersecurity outcome. Intelligence only matters when it moves, when it's enriched, prioritized, and actioned. Takeaway number eight. Threat intelligence isn't just defense. It's a business accelerator. Let's talk about the difference between metrics that look impressive and metrics that actually matter. Measuring activity without outcomes is like tracking meeting attendance. It shows who was in the room, but not whether anything useful happened. Here's what Justin Klein Keane of CSL Behring had to say about demonstrating value. Metrics can tell lots of stories and you're gonna lose control of your metrics as soon as you produce them, so you need to think really carefully about what kinds of data you want to produce. You want to stay away from data, that just shows volume or motion, but not direction or impact, and really sort of focus on the things that your program do does that adds direct value to the company. Keane made a powerful case. When cybersecurity teams align threat intelligence with real world outcomes like protecting revenue, brand equity, or uptime, they stop being seen by the c suite as a call center and start driving strategic value. Tobias Callas of SEB also spoke about the opportunity to create business value with threat intelligence. He stressed the importance of working with cross organization stakeholders from data scientists to subject matter experts and risk analysts to public relations teams, to understand their key security concerns and how those tie to business plans. Then he said, focus your intelligence on those specific contexts. There's often misconception when it comes to requirements that it's the the business, the stakeholders that should come with their requirements to you. My experience, however, tells me that they don't have time for that. And on top of that, they're not noticeable on what you can bring to the table. So this kind of engagement through a dialogue works perfectly. Excellent call out there by Tobias. Other departments may not understand the business value that cybersecurity brings, so be proactive about showing them how you can deliver value to their specific area. To sum it up, if your metrics don't show how security protects revenue, brand, or uptime, they're not business intelligence. They're just data points. Intelligence earns its seat at the table when it drives results, not just reports. Takeaway number nine. Track your adversary to protect your organization. Nation state actors don't announce their arrival. They probe silently, targeting edge devices and network infrastructure to establish footholds for espionage. This is exactly the behavior that red mic or salt typhoon uses. In this presentation, Sebas Scavaronelli adeptly summed up how the Insights group uses the Recorded Future platform, specifically network intelligence and malware intelligence, to track these advanced threats and translate them for our customers. Watch this. So if you don't know about malware intelligence, it is an AI assisted search tool that allows you to mine all the malware samples that Recorded Future has access to searching by malware family, by malware behaviors, through sigma rules, the arrows, super useful. On infrastructure side, we write all of those validated infrastructure detections. That means that we perform first hand Internet wise scanning to identify servers that are online that match known fingerprints of what a specific type of bad looks like. Once we have those detections in place, that's when the true positives that hit on that can automatically be pushed to the platform. And on the victim side, recorded feature network intelligence is definitely our best dataset for understanding what relactors are doing with those servers. How are they using them? What technologies are they targeting? What big things are they trying to compromise? The takeaway? Advanced threat actors rely on you not watching. Real time visibility and automated alerts turn their stealth into a liability because the moment something changes, you know. And that's exactly what the Insects Group and our platform deliver, visibility into adversary infrastructure and alerts when something new emerges. Takeaway number 10. Threats don't sleep and neither should your detection. As supply chains grow more complex, your attack surface can expand beyond your control. Adam Timmons of JPMorgan Chase shared how his team operationalizes threat intelligence on their third parties, embedding it into vendor workflows to catch compromises early and contain risk before it spreads. Take a look. Generally speaking, what we wanna do is really go wide with our more automated functions, and then we're trying to focus our manual work on the most critical stuff. We spent a lot of time and effort trying to figure out what's critical to the firm. And it really is to avoid that scenario where we find out a week into an incident, this company, we're sharing data with this company, and we didn't know. Timmins said that with over 6,000 suppliers, his team expands significant effort determining which suppliers are most critical and uses an algorithm to score the potential impact of a cyber incident at each critical supplier. To understand vendors that may be more vulnerable than others, the team overlays data from intelligence vendors like Recorded Future to guide their investigation. Now that kind of continuous intelligence monitoring isn't just critical for third party risk. It's essential for threat hunting across your environment, which brings us back to where we started, stopping threats automatically every single time. John Miller and Laura Hoffman from our team at Recorded Future shared a new blueprint for twenty four seven threat hunting. By combining human expertise with autonomous detection, teams can build always on hunting programs to actively seek out threats instead of waiting for alerts. Here's how autonomous threat operations from Recorded Future makes that possible. It reduces effort by continuously hunting threats. Set your parameters, specific indicators, malware, threat actors, and it investigates automatically with built in guidance. Ingesting intelligence from all your sources and prioritizing what matters. ISACs, commercial feeds, internal tools, everything gets correlated automatically. Enabling automated protection. When threat feeds are identified, validated malicious indicators get pushed directly into your security stack to block them in real time. Generating strategic reports using AI. Technical findings become strategic insights your leadership can understand and act on. I don't know who needs to hear this, but threats don't sleep. Your security teams do or at least they should. And yes, to all the analysts out there, that includes you. With automation, maybe you can finally get some actual rest and you can let the system pull the all nighters. And now we open it up for questions. If you haven't already, please write or type your questions into the chat. For those we are not able to answer live, our team will follow-up after the webinar. Yes. So it's looking like we have a lot of people really excited about autonomous threat operations. And the biggest question is when will it be available and how can I gain access? I can answer that. Yes. Autonomous for operations, it's gonna be fully launched in January 2026 with all the different capabilities that Adrian did walk through on stage. If you are a current customer and want to actually test out these capabilities, you can join our early access program, which you could sign up in the community, or you can talk to your customer success manager. Awesome. So seeing a question on more of, like, the MCP, AI workflow pieces, Is this something that other customers can do so very similar to what Grammarly did? Yeah. That's a great question, and they absolutely can do that. Grammarly showed Recorded Future in their own MCP that they built. But at Recorded Future, we're actually building out our own MCP that interested customers can now gain access to and sign up for through our community. So, please do that if you're interested, and feel free to connect with any of your customer success managers, for more details. Definitely. And I'll add a little bit more too. It's a pilot program, so you'll be reached out to by the product manager of AI as well as there's some of the agentic workflows too that I know will be released by the end of the year natively within Recorded Future AI. So be on the lookout for that for any of our current customers. Yeah. One more question I think we have time for here is, is there anything else that was released during predict that we did not have a chance to talk about here today? Oh, putting me on the spot. Yeah. So one thing I can think of is for all of our threat intelligence customers, we just released Auto Sigma. So Auto Sigma is available within Malware Intelligence. This is essentially a way to create custom Sigma rules based on the behavioral sandbox results that you can easily search in Malware Intelligence. Malware Intelligence, we launched back in April 2025, so this is a really exciting add on. And as mentioned, this is available today for all of our threat intelligence users. And as always, your feedback is extremely helpful, so please add feedback in the platform and reach out with any questions, especially in the Recorded Future community. Awesome. Thanks, Kathleen. And that's our show. Thank you so much for joining us. And if you do want to learn more about autonomous threat operations, please scan the QR code on your screen. See you next time.