The Security Strategist

Cybersecurity faces the continued onslaught of distributed denial-of-service (DDoS) attacks. Websites, applications, and online services flooded with junk traffic are unable to serve legitimate users. Businesses lose revenue, budgets are strained, and customers lose faith. DDoS may not get the attention of ransomware headlines, but attackers are changing their tactics to launch larger, more sophisticated attacks. In this day and age, it’s easier to orchestrate for a range of purposes, including extortion, disruption, hacktivism, or hurting competitors’ bottom lines.
Many are also powered by massive botnets made up of millions of compromised IoT devices. In this podcast episode of Security Strategist, host Richard Stiennon talks with Qrator Labs CTO Andrey Leskin about how these attacks are evolving and what organisations need to do to keep pace. They explore the growing scale and complexity of attacks, the role of massive botnets, practical approaches to DDoS mitigation, and how AI could accelerate existing attack capabilities.
The Biggest Trend is Scale
Leskin started at Qrator Labs as a developer 14 years ago and worked his way up to chief technology officer. “I started off as the guy in IT who woke up at 3 a.m. because something stopped working,” Leskin tells Stiennon. “Now I’m the lucky guy who gets to wake up and try to fix things for our clients.”
Today Qrator Labs manages cloud scrubbing infrastructure, bot management tools, and network monitoring services for hundreds of banks, betting platforms, e-commerce firms, media, education, tourism, and telcos throughout North and South America, Europe, the Middle East and Asia. That broad exposure gives him insight into the latest attack patterns. “Scale is the biggest trend,” Leskin says. An attack earlier this year topped two terabits per second and nearly one billion packets per second. It sustained that traffic rate for more than 40 minutes. During Q2, the company saw a doubling in terabit attacks (meaning attacks of one trillion bits per second or greater) year-over-year.
“That used to be a super-rare once-a-quarter type of thing,” Leskin said. “Twelve is not unique. Bandwidth that used to be exceptional is now just regular Tuesday.” Botnets powering these attacks are getting bigger, too. One botnet monitored by Qrator since March of last year grew from around 1.5 million bots to over 13 million within about a year. The geographic diversity of infected hosts also continues to expand, making filtering traffic based on region less effective as an automated mitigation technique.
Attacks are also easier to launch than ever before. Today attackers can find DDoS-for-hire services that simplify everything except deciding how much money they want to spend. Make the payment in cryptocurrency, paste in a target IP address or URL, and press launch. Many don’t need advanced technical knowledge. Decentralised command and control systems, including botnets using blockchain technology to coordinate activity, are complicating mitigation efforts further.
Existing Mitigations Fall Short
A common DDoS myth, Leskin says, is the idea that hosting with a cloud provider or CDN somehow provides adequate protection from DDoS attacks. While a website or app might remain available, those services are designed to maximise uptime and performance, not fend off attacks specifically. Organisations are still on the hook for all of the network resources an attack consumes. “And then when the monthly bill arrives you realise you were DDoS’ed on your wallet,” Leskin said.
Attackers are also leveraging multiple attack vectors more frequently. Instead of a single volumetric flood or application-layer attack, defenders might see both plus attempts to overwhelm other dependencies like a firm’s merchant processor. Leskin highlights how betting platforms saw an onslaught of attacks during the recent World Cup. Financial-services firms and fintech companies made up 44 per cent of DDoS attacks in Q1. That figure fell to 22 per cent in Q2 as attackers shifted their focus to gambling platforms, where attacks reached 1.5 terabits per second.
Tips for Defending Against Tomorrow’s Attacks
Preparing for these evolving threats starts with being operationally prepared, rather than buying into any one silver-bullet technology, Leskin said:
Know and understand your normal traffic profile down to the protocol level and by time of day or season. Traffic during a World Cup final will look very different to normal operations for a betting platform.
Expect blended attacks that use more than one vector designed to evade traditional DDoS mitigation systems.
Botnets are nothing new, but blocking them is still important. In the first quarter of 2026, Qrator blocked an average of 2.5 billion malicious bot requests each month. While not considered DDoS, these attacks can still have a significant impact on performance.
Have an incident response plan that you’ve practised so you can respond as quickly as possible when an attack happens.

From a tech perspective, there are two main categories of DDoS mitigation, each with advantages and disadvantages:
DNS-based protection
Easy to implement; works well at mitigating attacks against websites and web applications
Doesn’t work for everything routed outside of DNS, like voice services or game servers

BGP-based mitigation
Handles any type of network traffic at the network layer.
You need to own your own network; can take up to one full day to implement.

AI and DDoS Attacks
Leskin says that he doesn't expect AI to introduce new types of attacks. Instead, he sees it accelerating what already exists, helping attackers scan for vulnerable devices faster, automate reconnaissance, and grow botnets more efficiently. In other words, AI mostly lowers the cost and skill threshold for doing what attackers already do. Combined with the rise of DDoS-for-hire services, pushes more of the "easy attack" trend described earlier. His closing point was less about tools than posture: "Security isn't a state you achieve one time. It's a process you maintain, because whatever you're defending against is actively evolving against you."
The figures cited reflect Qrator Labs’ own network telemetry and provide a view into the attack trends observed across its protected infrastructure. While they do not represent the entire global DDoS landscape, they highlight a clear direction. For most organisations, the practical implication isn't "buy more bandwidth." It's building the muscle memory, traffic baselines, tested response plans, and mitigation that matches how you actually operate before an attack forces the issue. If you would like to learn more, visit qrator.net or follow Andrey Leskin on LinkedIn.
Takeaways
The scale and evolution of DDoS attacks from 2020 to 2026.
The role of botnets and their growth in size and geographic diversity.
Common motivations behind DDoS attacks.
Limitations of CDN and cloud provider protections against DDoS.
Best practices for organisations to assess and improve their DDoS resilience.
Technical mitigation techniques including DNS and BGP-based protections.
The importance of continuous security posture review.
Future trends including AI-driven attack methods and multi-vector incidents

Chapters
00:00 Introduction to the episode and guest Andrey Leskin
01:04 Overview of Qrator Labs and their cybersecurity services
02:46 The evolution and scale of DDoS attacks from 2020 to 2026
04:09 Growth of botnets and their geographic diversification
05:22 Motivations behind DDoS attacks and attacker profiles
07:49 Limitations of CDN and cloud protections against DDoS
09:16 Technical mitigation strategies: DNS and BGP protections
11:21 Proactive customer acquisition and security readiness
13:06 Key checklist items for organisations to improve resilience
17:24 Technical defences: DNS and BGP mitigation explained
21:07 Current threat landscape across industries and sectors
24:16 Future of DDoS attacks and AI-driven threats

The biggest cybersecurity challenges when it comes to integrating AI platforms like Microsoft Copilot, ChatGPT Enterprise or any other AI agents for enterprises may seem to be pertinent to governance, acceptable use policies and employee training in AI. However, that is not always the case. Ultimately, it comes down to a challenge with the data.
In the recent episode of The Security Strategist podcast, host Shubhangi Dua, Podcast Producer and B2B Tech Journalist, is joined by Itay Maor, Head of Product at Orion Security. They address the foundational issue with deploying agentic AI to enterprise workflows, which comes down to Data Loss Prevention (DLP).
Maor begins the conversation with the statement: “Data loss is preventable. It's not just observable.”
He adds that only by dropping assumptions built over 20 years of ineffective DLP can teams successfully make Data Loss Prevention work.
What Is Hindering Enterprise Security from Adapting to an AI-First World?
For enterprises to become a core part of an AI-first world, data must be protected from an early start. As soon as tools like Copilot or ChatGPT Enterprise enter the picture, sensitive data begins flowing into prompts. The issue is that security teams often lack visibility into what employees are inputting, such as customer records, deal terms, or source code.
Firstly, blocking the AI is not going to work in this scenario because AI is here to stay. The issue that needs addressing is that security teams need to be able to see where the enterprise data is flowing.
Maor believes AI hasn't created an entirely new security problem; it has exposed one that has existed for years.
“Customer records, source code, deal terms- legacy DLP doesn’t help much because they were built to match patterns, credit card numbers, keywords. Pasting a Q3 revenue forecast into a chatbot won't trigger standard security alerts,” he says, putting it into context.
“You approve ChatGPT Enterprise, but what if your employee just logged in using their personal account? Same URL, same interface, same browser, and your network controls say chatgpt.com and waves it through,” Maor adds.
Security teams need to know which identity the data is flowing to. Right now, it's difficult for them to differentiate.
The third layer, however, is where the market is heading because it depicts where the AI is connected to the data. For instance, Microsoft Copilot is wired into SharePoint and ChatGPT. Cloud connects to Google Drive, to Slack, and to email through native connectors. Meanwhile, the agents query internal systems on their own autonomously.
“There is no upload, no paste, no human action to inspect at all,” the Head of Product at Orion tells Dua.
AI agents end up inheriting 10 years of over-permisioning, he says; “it happily surfaces an M&A document to anyone with access that was never cleaned up, making it searchable in plain English.”
Each of these three layers widens the gap that all controls can cover. So, the first challenge isn't blocking AI; it's that you can no longer answer where your data is going, and everything else in AI security starts with that question.
How Security Teams Must Move From Detection to Data Loss Prevention
Maor proposes that enterprises need to shift their mindset from detection to prevention, asserting that "Prevention is the goal, not just detection with good reporting.
“Lead with the mindset before the tactics,” he advises enterprises, “data loss is preventable, and that should be the mindset, not just observable.”
This means security teams must stop enumerating every risk as a policy up front, unlike before. Policies are essential for deterministic rules, and they’re not going away. If a rule says ‘PCI data never leaves production’, but the era of managing hundreds of policies is over.
Another mindset shift is needed around false positives. Teams need to stop treating high false-positive rates as simply the cost of doing business. They're not some unavoidable force of nature. “They don't have to live with them. The problem is that when false positives become the norm, you train your team to ignore alerts—including the ones that actually matter,” he says to Dua.
And finally, enterprises need to stop staffing around the problem instead of solving it. Adding more analysts to a queue that's growing faster than your headcount isn't a scalable strategy. It's better to reduce the noise than to keep expanding the team that's trying to manage it.
As enterprises continue embracing AI, Orion's view is that the future of data security won't be defined by more dashboards or more point solutions. It will be defined by knowing where data is moving, understanding why it's moving and preventing loss before it happens.
Takeaways
DLP tools are overwhelmed with false positives.
AI can provide real-time contextual understanding.
Traditional DLP systems are not equipped for modern data challenges.
The future of data security relies on AI-driven solutions.
Guardrails are essential for safe AI usage in enterprises.
Real-time monitoring is crucial for effective data protection.
Policies should be limited and focused on specific use cases.
AI can recognise sensitive data patterns that traditional methods cannot.
Data security must adapt to the rapid evolution of AI technologies.
Education on new risks is vital for enterprises.

Chapters
00:00 The Evolution of Data Loss Prevention (DLP)
02:54 AI's Role in Redefining Data Security
06:12 Challenges of Traditional DLP Systems
09:02 The Need for Contextual Understanding in DLP
12:07 Guardrails for AI in Data Security
15:04 Transitioning from Policies to AI-Driven Solutions
17:54 Real-World Examples of Data Protection
20:49 The Future of DLP and Data Security
Watch the full episode of The Security Strategist podcast to hear Itay Maor, Head of Product at Orion, discuss how AI is reshaping enterprise DLP and what security leaders should act on next. Visit orionsec.io.
AI-native DLP, Data Loss Prevention, DLP, Enterprise DLP, AI Security, Enterprise AI Security, AI Data Security, Data Security, Microsoft Copilot, ChatGPT Enterprise, AI Agents, Agentic AI, Enterprise Data Protection, Sensitive Data, Cybersecurity, CISO, Contextual DLP, AI-Driven DLP, Data Loss Prevention AI, AI Security Strategy

You’re in an argument with your AI bot on ChatGPT; suddenly, your screen is locked. None of the keys on your keyboard work, and then you see a ransom note displayed on the screen. The systems have been encrypted, and the incident response team has been activated. The executives go to the one thing they had been told would save them, which is the backups.
Enterprises may believe their data is safe because of their immutable backups. But according to Mark Grazman, CEO of Fenix24, they are likely mistaken and often realise this after ransomware has already hit them.
At some stage of a ransomware attack, the assumptions of cybersecurity come up against reality.
In the recent episode of The Security Strategist podcast, host Richard Stiennon, Chief Research Analyst at IT-Harvest, is joined by Mark Grazman, CEO and Co-Founder of the ransomware recovery company Fenix24. They discuss the critical aspects of ransomware resiliency, including the four pillars of recoverability—survivability, completeness, speed, and assurance. They also talk about how enterprises can better prepare for and respond to attacks.
When Stiennon asked Grazman what's the thing he would assess that incident response playbooks miss if he walked into an active incident right now. Grazman says after a scoping call, he would ask the affected enterprise if their data was immutable. Most people say yes.
“There’s an 84 per cent chance that they’re wrong,” he adds. “The attack already happened, the data's already gone, and they don't even know it yet.”
The issue, he says that enterprises are practising and simulating that the data’s gone along with the infrastructure. “They're practising that there was a hurricane or a fire or a replication or an event as opposed to a true ransomware.”
Also Read: Ransomware Attacks: What You Need to Know
Find the latest cybersecurity insights, podcast episodes, and expert analysis on EM360Tech.cpm. Visit fenix24.com for more information.
Takeaways
84% of enterprises may be wrong about backup immutability.
Surviving backups do not guarantee successful recovery.
Ransomware recovery depends on four pillars: survivability, completeness, speed, and assurance.
Critical applications rely on more infrastructure than enterprises often realise.
Traditional disaster recovery tests may not reflect a ransomware attack.
Cybersecurity budgets need more investment in recovery readiness.

Chapters
00:00 Introduction to ransomware resiliency and Mark Grazman's expertise
01:20 Assessing incident response priorities in real-time attacks
02:06 The myth of immutable data and common misconceptions
03:06 Breaking down the four pillars of resiliency
04:03 Survivability: Protecting critical data and dependencies
05:02 Completeness: Ensuring full data and infrastructure recovery
07:32 Speed: Rehydration, containment, and infrastructure considerations
09:30 The importance of assurance and continuous testing
11:09 Applying resiliency principles to other disasters
12:37 The gap between enterprise expectations and reality
13:05 Evolving offence and defence in ransomware protection
14:39 Pre-attack preparedness and the Argos platform
16:06 The process of resiliency assessment and tuning
19:18 Organisational roles and collaboration for effective recovery
21:18 Key message for CISOs, CIOs, and CEOs on resiliency
23:30 Closing remarks and resources for further information

Security teams around the world have always tried to play a balancing act when it comes to authentication. If there are too many measures put in place, people will always find a way to get through it. In the world of rapid AI advancement, this balancing act is proving to be more difficult for organisations. The reason is that AI agents proliferate; they're now able to perform tasks on behalf of employees and customers without a human overseeing every action. So what needs to be done to prevent your organisation from being exposed?
On this episode of the Security Strategist Podcast, host Trisha Pillay talks with Dan Moore, Senior Director of CIAM Strategy and Identity Standards at FusionAuth, about how and why identity has become the new security perimeter. Moore has worked for almost six years at FusionAuth, starting in developer relations before stints in sales engineering and implementation prior to his current role. At FusionAuth, he helps track standards bodies like the IETF and OpenID Foundation and determines which fledgling methods are ready for adoption into the product.
The Security-Usability Tension Gets Sharper
Finding the right balance between strong security and a smooth user experience is a challenge organisations have faced for years. Moore traces it back to the invention of the first password field in the 1960s. Various industries have adopted different approaches to ensure that there is a balancing act of strong security and a smooth user experience for their customers. For example, banks are willing to require more security checks than a consumer app because the risks are so much higher.
The old methods of authentication were designed for a world where every login belonged to a person making decisions at human speed. This has all changed now because of AI agents. Unlike people, AI agents can work independently, run continuously, and complete thousands of tasks in seconds. This speed and scale mean they can also cause far more damage in a matter of seconds if something goes wrong. AI agents need to work independently, so traditional human-focused security measures like MFA and CAPTCHAs often get in the way. It's also important to know that removing those checks doesn't just eliminate the security risks. This simply means those risks can happen so much faster. At the same time, asking humans to approve everything isn't a solution either, because people quickly become overwhelmed and stop paying attention.
Adaptive Authentication in Practice
This is where identity is shifting from a single check at the door towards continuous and contextual verification. Moore describes it as moving away from a binary model, because risk no longer lives only at the login screen. It follows the session, the device and the ongoing behaviour within an application. FusionAuth worked with a platform connecting caregivers with families needing support, a sector handling sensitive data including that of minors. By adding enterprise single sign-on and multi-factor authentication, the company cut its authentication development time by 90 per cent and opened up business markets it previously couldn't serve, proof, Moore says, that tighter security and a better user experience aren't mutually exclusive when the approach is intelligent about context.
Giving AI Agents Their Own Identity
One of the biggest shifts discussed is the need to stop thinking of AI agents as just another user account. Instead, organisations need to manage them as separate digital identities with their own permissions and controls. Moore recounts a colleague mentioning they would let an AI assistant drive their browser while logged in as themselves. This becomes indistinguishable, from the system's perspective, from the person acting directly. Without a separate identity, there's no way to apply different policy, add extra checks, or restrict what an agent can do relative to its human counterpart. With all that said, it's no wonder that AI agents need their own identities, provisioning, and scope, along with their own audit trail. The risk comes down to velocity. A compromised employee can only do so much before they're detected, but a misbehaving AI agent can make thousands of decisions, access systems, and execute actions in the same amount of time.
Moore frames trust as resting on three interlocking layers: identity validation, audit, and policy enforcement. Validation establishes who or what is acting; audit records what actually happened, which matters given how unpredictable agent behaviour can be; and policy enforcement, built on principles like least privilege and short-lived, task-scoped credentials, limits the damage if something goes wrong. All three layers work together to build trust, he says, because each one compensates for what the others struggle to catch alone.
His advice for organisations still finding their footing is to start small rather than wait for a polished strategy: inventory the AI agents and automated processes already running, note what kind of credentials they rely on, and begin shifting static API keys towards short-lived, standardised grants. Above all, he argues, AI identities deserve their own category tied to an accountable human or team, but never simply reused from existing human or service accounts. If you would like to find out more about this, visit FusionAuth or connect with Moore on LinkedIn.
Takeaways
The changing role of identity in security.
Challenges of AI-powered applications and autonomous agents.
Adaptive authentication and risk-based security.
Building trust through identity validation, audit, and policy enforcement.
Practical steps for organisations to enhance security in AI environments.

Chapters
00:00 Introduction
01:28 Guest background and role at Fusion Auth
03:07 The security-usability tension in identity management
04:13 Impact of AI and autonomous agents on security
05:56 Balancing security controls with user experience
09:02 The shift to adaptive, context-aware authentication
11:48 Real-world example of security and usability balance
14:04 AI identities versus human identities
17:53 Building trust in AI systems with layered security
23:34 Practical steps for organisations to prepare for AI security
27:30 Closing remarks and resources

Prioritisation is the way to tackle enterprise data challenges. It may seem like a simple solution, and it might be too. If you’re an enterprise overwhelmed by vulnerabilities in data, especially with the evolution of AI and automation, this conversation is for you.
In the recent episode of The Security Strategist podcast, host Shubhangi Dua, Podcast Producer and B2B Tech Journalist at EM360Tech, sat down with James Walta, Vice President of Product Management at Brinqa. The agenda for this episode was to break down why enterprises are overwhelmed by vulnerability data. Additionally, Walta lays out a strategic plan of action to help enterprises prioritise vulnerabilities proactively rather than reactively.
The discussion builds on the previous episode, where Brinqa CSO Brad Hibbert and host Richard Stiennon, Chief Research Analyst at IT-Harvest, talked about how AI is helping attackers with faster scanning, smarter exploit chaining, and machine-speed intrusions.
Walta continues this conversation with EM360Tech’s Dua, focusing on prioritisation in exposure management strategies. He puts up a case noting AI will not rescue security teams from unorganisation unless the underlying data is ‘good’ and reliable.
Takeaways
Context is crucial for effective cybersecurity management.
The chaos in cybersecurity is amplified by AI-driven vulnerabilities.
Data quality is foundational for prioritisation and remediation.
Patching faster is not always the best approach; understanding risk is key.
Operational clarity can be achieved by unifying asset visibility.
Prioritisation must be based on business context and asset sensitivity.
AI can help but may also amplify confusion if data is poor.
CISOs should focus on outcome metrics rather than activity metrics.
Effective vulnerability management requires a clear understanding of ownership.
The conversation around cybersecurity must evolve to address real risk reduction.

Chapters
00:00 Navigating Cybersecurity Chaos
02:52 The Importance of Context in Cybersecurity
06:07 Bridging the Gap: From Vulnerability Detection to Remediation
09:09 Understanding Risk Over Speed
11:46 Enhancing Data Quality for Better Decision Making
14:57 Operational Clarity: Transforming Overload into Insight
18:05 Measuring Success Beyond Vulnerability Counts

Visit brinqa.com for more information on how enterprises should prioritise vulnerabilities proactively.
Vulnerability Management, Exposure Management, Cybersecurity Strategy, AI in Security, Risk Prioritisation, Brinqa, EM360Tech, The Security Strategist, Cyber Risk, Data Quality, CISO, Threat Exposure Management, Asset Visibility, IT Security, Risk Reduction, James Walta