Cogent Security has unveiled Cogent VR-1, an advanced reasoning model purpose-built for cybersecurity and designed to function within live enterprise environments. Alongside the launch, the company introduced IntrusionBench, a benchmark that evaluates AI models against realistic enterprise attack scenarios. On this benchmark, VR-1 demonstrated its ability to start from a single point of compromise, independently identify viable attack paths, and achieve roughly twice the performance of other leading AI models.

While many current frontier AI models are capable of identifying isolated software vulnerabilities, real-world cyberattacks are far more complex. Attackers typically exploit a sequence of seemingly insignificant weaknesses spread across different systems. A minor flaw in a public-facing application, excessive user privileges, or an unverified software artifact may appear harmless individually. However, when combined strategically, these weaknesses can provide a pathway to an organization's most valuable assets.

VR-1 was developed to emulate the investigative approach of experienced cybersecurity professionals. Beginning with an initial foothold inside an enterprise, the model analyzes the surrounding infrastructure, correlates vulnerabilities across multiple platforms, and determines whether a complete attack chain is realistically achievable. It can also evaluate whether proposed remediation measures genuinely eliminate the risk or merely shift it elsewhere. As a result, security investigations that previously required weeks of expert analysis can now be completed within hours.

The launch comes at a time when autonomous AI-driven cyberattacks are becoming an operational reality rather than a theoretical concern. Recent incidents have shown AI-powered attacks targeting major government institutions and technology companies worldwide. Organizations therefore require equally sophisticated defensive AI systems to identify potential attack paths before adversaries can exploit them and to strengthen their environments against increasingly capable AI-assisted threats.

According to Vineet Edupuganti, CEO of Cogent Security, many security vendors have promoted AI-enabled solutions over the past several years, but VR-1 introduces a measurable standard for advanced security reasoning. He explained that the model operates directly within a customer's environment to identify the attack paths that a highly capable adversary could realistically exploit and then helps validate effective mitigation strategies. Since attackers are already leveraging similar capabilities, the company's objective is to ensure defenders have access to equivalent technology within a controlled and well-governed framework.

Addressing the Limitations of Existing AI Models
Traditional AI security benchmarks primarily assess whether a model can exploit an individual vulnerability in isolation. IntrusionBench instead evaluates the model's ability to conduct realistic, multi-stage enterprise attacks. In the benchmark, the AI agent begins with limited access and a defined objective, then attempts to navigate through cloud infrastructure, identity management systems, and internal enterprise applications to reach a specified target. Performance is measured based on successfully executing verified attack chains rather than simply identifying theoretical possibilities.

Developed by Cogent Security, IntrusionBench demonstrated that VR-1 performs particularly well when operating with minimal prior knowledge of the target environment. As progressively more environmental information becomes available, competing models improve and the performance gap narrows. Once the complete source code and underlying vulnerabilities are disclosed, the models perform at similar levels, suggesting that VR-1's primary strength lies in investigating uncertainty and connecting fragmented information rather than merely exploiting known weaknesses.

Geng Sng, Co-founder and CTO of Cogent Security, noted that existing frontier AI models can identify attack paths largely because of their general coding and reasoning abilities, rather than being specifically engineered for cybersecurity. VR-1, by contrast, has been trained explicitly to understand the complex, multi-stage attack sequences commonly encountered in enterprise environments, enabling it to outperform general-purpose AI models on practical defensive security challenges.

Secure Runtime Environment
In addition to VR-1, Cogent introduced the Cogent AI Harness, a runtime environment designed to safely deploy AI reasoning models within operational enterprise networks. The platform provides controlled access to environmental context, policy-aware tools, governance mechanisms, and independent verification processes. Every action performed by the AI is validated against organizational security policies and supported with verifiable evidence.

Cogent follows a validation philosophy described as "prove, don't detonate," meaning that vulnerabilities are confirmed without causing operational disruption, establishing persistence, or damaging production systems.

Controlled Availability
Recognizing that advanced cybersecurity capabilities could also be misused, Cogent has chosen not to release VR-1 as an open model. Instead, access is restricted to vetted organizations through the Cogent Frontier Access Program, where customers receive the model with built-in safety controls, governance policies, and comprehensive audit logging.

Organizations accepted into the program collaborate directly with Cogent's research team to evaluate the model, validate its performance within their own environments, and deploy it securely. Eligible participants can also receive a Frontier Model Risk Assessment, which identifies the attack paths an advanced AI model could realistically exploit, highlights vulnerabilities that become more significant as AI capabilities evolve, and recommends opportunities to safely leverage AI to strengthen organizational security.

Organizations can apply here.

ISC2, the world's largest nonprofit association for cybersecurity professionals, has announced the creation of a new certification focused on artificial intelligence (AI) security. As part of the initiative, the organization is inviting cybersecurity practitioners worldwide to volunteer their expertise in defining the knowledge, skills and competencies required to secure AI technologies, manage AI-related threats and reduce associated risks. Developed using ISC2's practitioner-driven certification methodology, the new credential is intended to become a globally recognized, vendor-neutral benchmark for AI security expertise.

With AI being rapidly integrated into enterprise operations and cybersecurity workflows, security professionals are increasingly expected to safeguard AI-powered systems. Despite this growing responsibility, the industry currently lacks an internationally accepted certification that validates AI security capabilities.

"The cybersecurity community has made it clear that there is a strong need for practical and credible ways to demonstrate AI security expertise," said Casey Marks, Ph.D., Chief Operating Officer of ISC2. "This certification is being created to address that need by helping professionals and employers verify AI security competencies, strengthen the cybersecurity workforce and increase confidence in organizations' ability to protect AI-driven environments."

Standards-Based Development
As AI continues to reshape cybersecurity operations—from reviewing AI-generated outputs and supervising automated systems to making critical risk management decisions—ISC2 is assembling cybersecurity professionals from across the globe to help establish a vendor-neutral certification framework.

The development process will identify how traditional cybersecurity principles are being applied within AI-enabled environments. Areas under evaluation include AI risk management, secure-by-design practices, data security, governance, oversight and other essential security disciplines. Input from experienced practitioners will determine the tasks, knowledge domains and practical skills that will form the basis of the certification and its future examination.

Industry Participation
ISC2 is encouraging cybersecurity professionals, whether members or non-members, to participate voluntarily in the certification's development. Their contributions will help ensure the credential accurately reflects current industry practices and the competencies organizations expect from AI security professionals.

Volunteers will have opportunities to:

Marks emphasized that practitioner involvement will play a key role in establishing a trusted global standard for AI security, enabling organizations to better identify and validate professionals with specialized expertise in securing AI systems.

Individuals interested in contributing can find additional information and register their interest through the ISC2 website.

Development Timeline
ISC2 has already begun the initial planning and definition phase of the certification. This will be followed by content development and a pilot examination later in the year. The organization plans to provide regular updates on project milestones, opportunities for community participation and certification progress. The pilot version of the AI Security certification is expected to be available by the end of 2026.

AI Spotlight Event
To further promote discussion on the intersection of AI and cybersecurity, ISC2 is organizing a two-day virtual "ISC2 Spotlight: Artificial Intelligence" event on July 14–15. During the event, Thomas Jackson, CISSP, Senior Exam Content Development Manager, and Willard Smith, CISSP, Exam Content Developer, will present a session titled "The AI Pivot: Safeguarding the Future of ISC2 Certifications." Scheduled from 9:00 a.m. to 9:45 a.m. EDT on July 15, 2026, the session will provide an overview of the AI security certification initiative and explain how cybersecurity professionals can participate in its development.

ISC2 Members can register here for this event, which is totally free of any charge.

Click here to know more about the ISC2 AI security certification program and volunteer opportunities.

Silent Push, a company specializing in proactive cyber defense, has unveiled Silent Push 6.0, a significant upgrade to its cyber defense platform. The latest version is designed to help security teams act more quickly and efficiently through improved usability, enhanced visibility into threat infrastructure, and expanded integration capabilities. The release introduces upgrades to Traffic Origin and new workflow features that support a threat-informed security strategy focused on identifying and mitigating risks before attacks occur.

“Silent Push 6.0 is designed to cut through the noise and deliver actionable intelligence on adversary infrastructure before attacks materialize—something unmatched in the industry,” said Ken Bagnall, CEO and Co-Founder of Silent Push. “This release empowers enterprises and government organizations to identify emerging threats sooner, enabling informed decisions that strengthen cybersecurity defenses at the earliest stages of the attack lifecycle.”

The new platform enhancement is built to integrate seamlessly into existing security operations, helping teams focus on high-priority threats without requiring additional tools or major workflow changes.

Key Enhancements in Silent Push 6.0
Traffic Origin
A newly developed capability exclusive to Silent Push, Traffic Origin provides organizations with deeper visibility into malicious activity, infrastructure shifts, and traffic behavior to accurately identify the true geographic source of internet activity. The feature includes a dedicated user interface, access to a full year of historical data, and more granular permission controls. It supports a range of use cases, including fraud detection, Know Your Customer (KYC) verification, recruitment validation, and incident response investigations.

MCP Server Integration
Silent Push has introduced a hosted endpoint that allows its threat intelligence data to connect directly with AI platforms such as Claude, ChatGPT, and other large language models. Security analysts can conduct investigations using natural language queries, enabling tasks such as indicator enrichment, historical DNS analysis, infrastructure mapping, adversary clustering, risk assessment, and creation of investigative reports. By reducing unnecessary context and streamlining data access, the MCP Server helps accelerate investigations while lowering AI token consumption.

Redesigned Navigation and Module Structure
The platform now features three primary modules—Defend, Insight, and Reconnaissance—to simplify navigation and improve operational efficiency. Defend focuses on proactive integrations with SIEM, SOAR, firewall solutions, and Silent Push’s proprietary Indicators of Future Attack® (IOFA). Insight delivers comprehensive artifact context within a unified interface to speed up threat triage. Reconnaissance supports threat hunting and incident response teams by enabling infrastructure tracking and analysis. An additional module, Advanced Attribution, serves as the home for the Traffic Origin capability.

Enhanced TLP Amber Reports
Backend and infrastructure improvements have significantly increased report-loading performance. The updated reporting framework now supports richer content formats, including embedded video tutorials accessible directly within reports. Threat indicators referenced in reports can be explored further through pivoting capabilities, while the newly available TLP Amber Report API provides programmatic access to report metadata and related information.

Bulk Enrichment in Insight
The Bulk Enrichment feature enables analysts to evaluate multiple indicators simultaneously within a centralized dashboard. Designed for the early stages of investigations, it helps security teams quickly prioritize suspicious assets by providing immediate access to Silent Push intelligence. This capability addresses limitations commonly found in single-indicator lookup tools, allowing analysts to identify the most critical threats faster and more effectively.

Click here to know more.

Small Businesses in a Digital World: Opportunities, Challenges, and Cybersecurity
Today, small businesses are anything but small in their impact. They have long been the backbone of the global economy, and with the rise of digital technology, they are expanding beyond local communities into international markets.

However, this increased digital presence comes with both opportunities and risks. Many small business owners, who once believed they were not targets for cybercriminals, are now facing the harsh reality of cyber threats firsthand.

A soon-to-be-released Mastercard survey of over 5,000 small and medium-sized business (SMB) owners across four continents found that 46% have been victims of cyberattacks. Alarmingly, nearly one in five of those affected were forced to shut down or file for bankruptcy following an attack.

The most common types of cyber threats—hacking, malware, and phishing—inflict significant financial damage. Business operations slow down, revenue is impacted, and reputational harm can last long after the breach. In fact, 80% of affected business owners reported needing to rebuild trust with customers and partners post-attack.

Recognizing the growing risks, cybersecurity is now a top priority for small businesses. According to the survey, 86% of SMBs have conducted cybersecurity risk assessments and implemented security plans. However, only 23% feel highly satisfied with their preparedness, and an equal percentage feel confident in detecting threats. In other words, many business owners recognize the gaps in their knowledge.

Unfortunately, cybercriminals are well aware of these vulnerabilities. With limited budgets and resources, small business owners often juggle multiple roles—including managing their own IT security. One of the biggest challenges they face is ensuring employees take cybersecurity seriously. According to the survey, 73% of business owners struggle to get staff to follow best practices, and only a quarter are confident in their ability to educate employees on cybersecurity.

To strengthen their defenses, small businesses need support. That’s why Mastercard is committed to helping them securely manage payments, access funding, and embrace digital tools while protecting themselves from cyber threats. Since 2020, we’ve helped integrate over 50 million small businesses into the digital economy—but trust is what keeps them there.

As part of our commitment, we are collaborating with cybersecurity experts and investing in solutions that simplify security while enhancing protection. For instance, our My Cyber Risk tool adapts the RiskRecon vulnerability scanner to help small businesses identify and address security threats on their websites and apps.

Additionally, the Mastercard Business Builder program, launched in February, provides credit and debit cards designed specifically for small businesses, bundled with cybersecurity services to safeguard against fraud. One key feature, ID Theft Protection, offers monitoring, alerts, and resolution support to defend against identity fraud.

While identity theft is often associated with individuals, businesses are also at risk. Fraudsters can misuse business credentials to apply for loans, file fraudulent tax returns, or even claim ownership of intellectual property. To combat this, Mastercard is leveraging open finance tools to verify account ownership, authenticate users, and secure financial transactions.

We are also expanding cybersecurity solutions through strategic partnerships. By teaming up with global cyber firm VikingCloud, we combine Mastercard’s advanced risk evaluation tools with VikingCloud’s remediation solutions, benefiting millions of merchants worldwide. Similarly, our partnership with CyberMonks offers a marketplace where businesses can easily find and implement customized cybersecurity strategies.

As technology advances and cyber threats evolve, continuous education is essential. The Mastercard Trust Center provides small business owners with free access to cybersecurity research, tools, and training, developed in collaboration with the Global Cyber Alliance, Cyber Readiness Institute, and National Cybersecurity Alliance.

By working together, we can create a safer digital environment for small businesses. After all, security isn’t just about defense—it’s about fostering trust and enabling growth. With the right protections in place, small businesses can operate with confidence and seize new opportunities without fear.