Archive for May, 2026

« Older Entries
Newer Entries »

U.S. employers are falling behind their own workforce on AI 

Posted in Commentary with tags on May 27, 2026 by itnerd

Nexthink has issued new analysis showing that employer support for AI is lagging real-world U.S. workforce adoption. Drawing on data from Gallup, the Federal Reserve Bank of New York, JFF, and Forrester – combined with Nexthink usage data from millions of endpoints – the findings show AI adoption has become a game of chance, with employees left to navigate tools without support or guidance.

According to Gallup, 28% of U.S. employees now use AI at work at least a few times a week. Yet research from the Federal Reserve Bank of New York shows just 15.9% of workers say their employer currently offers any AI training – a gap that makes clear employer support is failing to keep pace with AI usage. Nearly six in ten workers who consider AI training important are not being offered it, with the New York Fed finding demand for training (38%) more than double the share of employers providing it.

Despite this, JFF research shows 56% of workers have not been consulted by employers on how AI tools are used in their work. And when they seek guidance, workers turn to social media (31%), news articles (27%), or friends and family (21%) rather than employers (9%).

The scale of unsupported AI use is already visible. Nexthink data, drawn from 4.9 million sessions per day across 3.4 million employees, shows GenAI users engaging with these tools an average of 10 times a day and spending three hours and 14 minutes per week doing so. With adoption at this level occurring without formal guidance, the window for employers to get ahead of adoption is narrowing fast.

The challenge will only become more pronounced as AI becomes a larger part of everyday work. Forrester projects that AI will augment 20% of jobs over the next five years, raising the stakes for employers to understand not only whether AI tools are being used, but whether employees have the support, training and digital experience to use them effectively.

Leave a comment »

Detectify launches MCP Server to secure the autonomous coding loop

Posted in Commentary with tags on May 27, 2026 by itnerd

Detectify has announced the launch of the Detectify MCP (Model Context Protocol) Server, a new integration layer that brings Detectify’s security testing engines directly into AI-driven development workflows, helping coding agents find and validate exploitable vulnerabilities and interpret attack surface data with unprecedented precision.

As organizations increasingly rely on AI agents to write, refactor, and modernize code, software production is accelerating faster than many security teams can realistically review or govern. Whether through official engineering initiatives or shadow AI adoption by developers eager to speed up work, AI-assisted development can help eliminate some common coding mistakes. It is also dramatically increasing the volume of software, services, APIs, and infrastructure organizations must continuously track and secure. The result is a growing visibility and governance challenge, compounded by shadow IT and shadow AI adoption, where organizations may write cleaner code while simultaneously losing track of what they are deploying and exposing.

The Detectify MCP Server is designed to close that gap by giving AI agents a standardized way to augment development and security workflows with AI-assisted access to Detectify findings and capabilities, allowing them to access and act on real-time security findings as part of autonomous development workflows. Rather than relying on periodic reviews or delayed security handoffs, organizations can embed continuous validation more directly into the software delivery process as code, infrastructure, and services evolve.

Key MCP Server capabilities include:

  • “Find & Fix” Automation: Instead of security findings landing in a static backlog, they can now be handed directly to AI agents as structured remediation tasks. Agents can generate a patch, trigger a Detectify validation scan to confirm the vulnerability is resolved, and present a verified fix for human review.
  • Conversational Command: Query scan results, monitor asset status, and surface high-severity findings through natural-language interactions connected to the Detectify MCP Server.
  • Frictionless Setup: A lightweight configuration allows organizations to connect their preferred AI tools to the remotely hosted Detectify MCP server for simplified deployment and connectivity.

Traditional application security workflows were built around slower development cycles, where human review and periodic testing could reasonably keep pace with software delivery. In modern AI-assisted environments, those assumptions are increasingly breaking down as code, infrastructure, and services evolve continuously.

The launch reflects a broader shift in AppSec toward continuous, real-time security validation. While LLMs excel at reasoning, frontier models operate at a speed and cost-profile that makes large-scale security testing impossible. Detectify solves this by monitoring millions of changing domains using compiled, deterministic code, while the MCP Server combines that scale with agentic workflows to help security operate at the same velocity as engineering.

As AI-assisted development continues to accelerate engineering velocity, organizations face increasing pressure to move beyond one-time security reviews and maintain continuous visibility into what exists across their attack surface.

The Detectify MCP Server will be available soon as part of Detectify’s continued investment in AI-native application security. For more information, visit Detectify.com.

Leave a comment »

AI Adoption Creates Critical Cloud Security Gaps for Enterprises

Posted in Commentary with tags on May 27, 2026 by itnerd

Check Point has released its 2026 Cloud Security Report: Enter the AI Era, revealing a growing disconnect between rapid AI adoption and security readiness.

The report reveals a critical shift from the cloud “blind spots” of 2025 to a deeper challenge in 2026: organizations are no longer just struggling with visibility, but with governance, control, and real-time enforcement. AI is changing how users behave, how applications communicate, and where threats enter the environment. This year, 77% of organizations have updated their security strategy for cloud in response to AI, yet only 26% report having the architecture to enforce it. This reveals a 51-point gap between intent and capability.

Meanwhile, attackers are weaponizing AI tools to accelerate phishing, generate malware, and launch adversarial attacks faster than traditional security models can respond. The impact is already measurable: 78% of organizations reported confirmed or suspected AI-related security incidents over the past year.

Key findings for cloud-native environments include:

  • Infrastructure Misalignment: 52% of AI workloads span hybrid environments, yet 64% say their architecture needs redesign
  • Perimeter Gaps: 76% rate datacenter security as critical for AI, but only 35% say it can support current needs
  • Performance Challenges: Only 24% can fully inspect AI traffic without impacting performance; 71% report increased WAF false positives
  • Operational Complexity: 88% say AI has increased security complexity; 67% report fragmented policies
  • Limited Visibility: 54% of organizations have experienced an AI-related security incident, while another 24% cannot confirm due to lack of visibility. This means more than three-quarters have either been hit or cannot determine whether they have
  • Identity Risks: 48% cite non-human identities (AI agents, APIs) as a top concern
  • Inconsistent access model: Organizations have yet to converge on a single access model. 24% say they have no AI-specific access controls, and only 16% enforce controls consistently across the environment

Closing the AI Security Gap

To address these challenges, the report emphasizes the need for a unified, prevention-first architecture across cloud, datacenter, SaaS, and endpoints.

Check Point’s Hybrid Mesh Network Security approach delivers:

  1. Unified Management: 86% of leaders rate unified security management across cloud, datacenter, and edge as critical for AI workloads. A hybrid mesh architecture keeps policies and protections consistent everywhere, no matter where data or workloads run
  2. Prevention-First Security: Real-time blocking of ransomware, zero-day threats, and data leaks using AI-driven insights, validated by a 99.8% security effectiveness score in the 2026 Miercom report
  3. Secure Connectivity and Threat Prevention: Identity-based protection ensures every user, device, and application is verified and protected in real time, with consistent security across all access points and without impacting performance
  4. AI Defense Plane: A unified control plane governing how AI is connected, deployed, and operated, with runtime protection across employee AI use, applications, and agentic systems
  5. Agentic Network Security Orchestration: The 51-point enforcement gap is more than a visibility problem; it’s also an operational one. Check Point’s newly launched Agentic Network Security Orchestration Platform shifts security teams to the level of business intent, letting AI agents autonomously handle policy creation, Zero Trust tightening, and compliance across hybrid environments

Download the full 2026 Cloud Security Report: Enter the AI Era here, or read the accompanying blog post.

Leave a comment »

Claud Mythos Threat Report By BforeAI’s PreCrime Labs

Posted in Commentary with tags on May 27, 2026 by itnerd

In the six weeks following Anthropic’s Mythos announcement, BforeAI’s PreCrime Labs identified 3,188 domains built to exploit the Claude and Mythos brands. These weren’t simple typosquat pages – they were enterprise-style platforms impersonating AI security scanners, developer tools, vulnerability assessment services, and account marketplaces. Their targets: developers, security teams, and organizations curious enough about Mythos to hand over API tokens, infrastructure data, or payment credentials without realizing the platform wasn’t legitimate.

This series of threats are coming after Anthropic’s recent announcement of Mythos, and its shareability to a limited set of vendors to find vulnerabilities.

You can read more here: https://bfore.ai/report/anthropic-mythos-phishing-domains-threats-exploiting-claude/

Leave a comment »

Iranian hackers responsible for LA transit system breach, Israeli researchers say

Posted in Commentary with tags on May 27, 2026 by itnerd

It is being reported that Iranian hackers were responsible for a disruptive computer breach in March that forced Los Angeles’ transit system to shut down parts of its network.

More details can be found here: https://gambit.security/blog-posts/babil-of-minab-iran-mois-destruction-campaign

Commenting on this news is Ensar Seker, CISO at SOCRadar: 

“This incident reflects a broader shift we are seeing in Iranian cyber operations: the growing willingness to combine espionage, disruption, and psychological impact in a single campaign. Transportation systems are particularly attractive targets because even limited operational disruption can generate immediate public visibility, media attention, and pressure on local governments. In this case, the theft of hundreds of gigabytes of internal data alongside network disruption suggests the attackers were not simply conducting intelligence collection, but also positioning themselves for coercive influence and operational impact.

What is especially concerning is the targeting profile. Public transit environments are highly interconnected ecosystems that depend on legacy infrastructure, third-party vendors, operational technology, and real-time communications systems. That creates multiple attack paths for adversaries linked to state-sponsored ecosystems such as Iran’s MOIS-affiliated actors. Even if attackers do not directly impact train operations or safety systems, disruption to scheduling, internal communications, identity systems, or maintenance platforms can still create significant operational paralysis.

Organizations should also pay attention to the data exposure aspect of this incident. The theft of backups, emails, and internal documentation can create long-term downstream risks including follow-on phishing campaigns, extortion attempts, infrastructure mapping, and targeting of employees or contractors. Many organizations still treat operational disruption and data theft as separate problems, but modern state-aligned actors increasingly combine both into multi-stage campaigns.

This attack also reinforces an important geopolitical reality: regional conflicts increasingly spill into civilian digital infrastructure outside the immediate conflict zone. Transportation, healthcare, energy, and municipal services are becoming symbolic and strategic targets for adversaries seeking asymmetric pressure without crossing traditional military thresholds.”

The ability to set up shop and conduct activities that takes weeks and months isn’t good. Thus it should be one more thing that organizations should watch out for when conducting counter surveillance.

Leave a comment »

Over half of website searches end without action, new data reveals

Posted in Commentary with tags on May 27, 2026 by itnerd

New research from Finnish AI-powered search and content discovery company AddSearch reveals that more than half of all website searches (58.3%) end without users taking further action. Based on an analysis of over 337,000 real-world search queries across 11 university websites between January and April 2026, the findings suggest that traditional website search experiences are increasingly failing users at the exact moment their intent is highest. The analysis exposes a widening disconnect between how people now seek information online and how most websites are still designed to deliver it.

A further 16.1% of searches returned no purposeful results at all,  meaning roughly one in six users who actively turned to a site’s search bar found nothing relevant in response. Taken together, the data exposes a structural problem affecting content-heavy websites across sectors: the queries people are submitting have outpaced the ability of traditional search infrastructure to respond to them. Users are searching in natural language, asking complete questions, and expecting immediate direction,  while most website search experiences still return ranked lists of links, leaving interpretation entirely to the user.

The findings reflect a broader behavioural shift already reshaping digital experiences across industries. As generative AI tools such as ChatGPT and Gemini become embedded in everyday online behaviour, users increasingly expect websites to provide direct answers, understand conversational language, and reduce the effort required to find information. Traditional keyword-based navigation patterns are giving way to more natural, intent-driven interactions.

The analysis excluded empty searches, bot traffic, malformed queries, and spam injections in order to focus exclusively on meaningful user interactions. In total, the dataset covered 337,799 searches, offering a large-scale snapshot of how users interact with modern site search experiences in high-information environments.

The search queries analysed were overwhelmingly high-intent actions tied to important decisions and tasks. Across the universities studied, the most common searches included degree programmes such as nursing, business, psychology, and engineering, alongside administrative tasks including financial aid, transcripts, applications, graduation requirements, and accommodation. These are not casual browsing behaviours. In many cases, they represent moments where prospective students are evaluating major life decisions or existing students are attempting to complete urgent administrative tasks. Yet despite that intent, nearly two in three searches ended without further engagement.

According to the analysis, the problem is rarely a lack of content. Universities already publish extensive information about programmes, applications, financial aid, and student services. Instead, the issue lies in how that information is surfaced and delivered. Traditional site search systems were designed primarily for browsing and retrieval, presenting users with lists of links and leaving the interpretive work entirely to them. In contrast, modern users increasingly expect websites to interpret intent, summarise relevant information, and guide them toward next steps. 

The gap becomes especially visible in natural language searches, long-tail queries, and spelling variations, many of which traditional keyword-based systems struggle to interpret effectively. A user searching for “accommodation options for international students,” for example, may receive no results despite the relevant information existing elsewhere on the site. The research suggests these are no longer edge cases, but an increasingly common reflection of how people naturally interact with digital systems, phrasing queries the same way they would ask another person or an AI assistant.

The implications extend beyond user experience metrics. In higher education environments, failed searches for programs, applications, or financial aid can signal that prospective students are abandoning key decision-making journeys. Across other industries, including healthcare, financial services, e-commerce, government, and media, every unsuccessful search increases friction at a critical moment of intent, contributing to lost engagement, lower conversion rates, and rising support demand. 

AddSearch’s analysis, based on the dataset studied, indicates that AI-powered search layers could directly improve outcomes for between 65% and 75% of all search interactions, covering programme and degree queries, administrative task searches, and the 16.1% of queries currently returning no results. Under a conservative scenario, reducing the no-click rate from 58.3% to 40% and recovering 30% of no-hit searches, the projected impact is a 20-30% increase in meaningful engagement per search session, alongside faster task completion and a measurably reduced inbound support volume.

Leave a comment »

Customer Satisfaction Now the #1 Metric Improved by AI Agents in Canadian Service Organizations

Posted in Commentary with tags on May 27, 2026 by itnerd

Here are findings from the State of Service: AI Agents Edition. The report is based on a survey of 3,075 customer service professionals globally, including 150 respondents from Canada.

A year ago, service organizations were debating how AI agents might deliver, especially in Canada where the market has historically lagged in adoption. This year’s findings show adoption is up, with Customer Satisfaction and Average Handle Time tied for the #1 improved metric, ahead of traditional operational metrics like service rep productivity, customer retention, and first response time.

Top Canadian takeaways we believe will resonate with your audience:

  • Agentic AI adoption is accelerating across service organizations: 55% of Canadian service organizations currently use agentic AI.
  • Customer trust and transparency remain critical: 78% of organizations allow customers to request a human rep during an AI interaction, while 50% disclose when customers are interacting with AI.
  • Organizations are seeing fast time-to-value from AI agents: 65% of organizations report seeing value from AI service agents within 31–60 days, while another 20% see value within 30 days.
  • Expanded Agentic AI Use: 73% of Canadian service professionals say their organizations would benefit from expanding their AI agent use. 
  • A Budget Priority: In fact, 60% of Canadian service professionals say “customer-facing AI” is their top AI budget priority in the next 6 months.
    • 36% are increasing their AI budget significantly this year 

You can read the report here: http://salesforce.com/news/stories/ai-service-agents-improve-customer-satisfaction

Leave a comment »

RevEng.AI raises $15 million to verify the security and integrity of AI-generated software

Posted in Commentary with tags on May 27, 2026 by itnerd

RevEng.AI today announced it has raised a $15 million Series A round led by NATO Innovation Fund with participation from Sands Capital, In-Q-Tel (IQT), IQ Capital, and Episode One.

RevEng.AI helps organizations analyze compiled software at the binary level to determine what is actually inside executables, firmware, and third-party software without requiring access to source code. While Anthropic’s Mythos has drawn significant attention to frontier AI for cybersecurity and autonomous vulnerability discovery, RevEng’s foundational AI model named BinNet translates these advances into practical application by working directly on released software binaries to find novel cyber threats in real-world environments. It has been trained with elite cyber units inside allied governments and top commercial teams to automatically hunt down hidden security vulnerabilities and backdoors placed in widely used software. This enables the platform to automate the analysis of binary software, identifying hidden threats, malicious functionality, unexpected components, vulnerabilities, and abnormal release changes before software is shipped, purchased, or deployed.

The company aims to use this technology to address one of the largest unsolved problems in the cybersecurity space: software supply chain integrity. Software supply chain attacks have become one of the fastest-growing and most consequential risks facing enterprises, governments, and critical infrastructure operators. Every organization now depends on software it did not fully build or inspect, including commercial applications, open-source dependencies, firmware, and vendor updates. Once that software is compiled into machine-executable form, its contents are no longer easily visible to humans—creating a powerful opportunity for attackers and a major blind spot for defenders.

That blind spot has become far more urgent as AI automates software development. In the last year, coding assistants have evolved into AI coding agents that autonomously generate, modify, and deploy increasingly large amounts of software. As a result, organizations face a new bottleneck in verifying the artifact that ships and ensuring it has no malicious or high-risk functionality embedded inside. RevEng.AI is built to solve this problem.

RevEng combines proprietary binary analytics with AI-assisted reasoning to automate work that has historically required scarce, highly specialized reverse engineering expertise. This allows organizations to analyze software at a fraction of the cost and time of manual review while scaling integrity checks across the complete binary artifacts they depend on.

Unlike legacy application security tools, which primarily inspect source code, repositories, or declared package metadata, RevEng verifies the compiled executables and artifacts. The platform works directly on closed-source, third-party, and internally developed software, enabling organizations to:

  • Identify hidden, undeclared, or unexpected components in binaries and executables
  • Detect malicious functionality, dangerous behaviors, and vulnerabilities that may not be visible in source-centric workflows
  • Compare new releases against trusted versions to uncover abnormal drift or suspicious changes
  • Make trusted decisions before software is shipped, procured, deployed, or updated

The company is already seeing strong early demand with enterprise and defense organizations who are working with and investing in the solution. RevEng integrates into existing security and software delivery workflows, helping customers move from reactive investigation to proactive software verification, stopping software supply chain attacks in their tracks.

For more information, visit: https://reveng.ai/

Leave a comment »

Hackers claim OnlyFans leaked 340 million user records

Posted in Commentary with tags on May 27, 2026 by itnerd

On a popular data leak forum, hackers claim to be selling 340 million OnlyFans user records, including emails, usernames, and account activity metrics. According to Hackread, the attackers claim they built the database using data from previous OnlyFans leaks, public sources, and other data breaches.

The Cybernews research team has investigated the data sample and here’s what they found:

  • There were only 10 sample records attached to the post.
  • The listed data included user IDs, usernames, email addresses, and registration profiles. Other fields in the sample that were empty were for phone numbers, account flags, and linked accounts.
  • The team noted that samples in the account appear to come from around August 2025, which could indicate the attacker collected older data.

For more information, here’s the full report: https://cybernews.com/security/onlyfans-mega-data-leak-hackers-claim/ 

Leave a comment »

Hisense Partners with FIFA for First-Ever Sensory-Inclusive FIFA World Cup

Posted in Commentary with tags on May 22, 2026 by itnerd

Hisense today announced a groundbreaking partnership with FIFA and KultureCity to support the first-ever Sensory Inclusive tournament at the FIFA World Cup 2026™.

Through this initiative, all 16 host stadiums across the United States, Canada and Mexico will feature dedicated sensory rooms equipped with Hisense display technology. Designed for fans who experience sensory overload — including individuals with autism, PTSD, dementia, anxiety and other conditions — these spaces will provide calming, supportive environments within the high-energy setting of match day.

Expanding Access to the Beautiful Game

Research indicates that an estimated five per cent to 16.5 per cent of people experience sensory processing challenges. For these fans, the intensity of live sporting events — the high energy of the crowd, sudden cheers and ongoing movement — can make attending feel overwhelming or inaccessible. This initiative looks to change that, ensuring that more fans can experience the beautiful game in person.

Sensory-Inclusive Infrastructure Across All 16 Stadiums

The initiative centers on two key components:

  • Sensory Rooms at Every Stadium: Each of the 16 venues will include a dedicated quiet space where fans can step away from match intensity to regulate their sensory experience. These rooms will feature dimmed lighting, reduced noise, comfortable seating, tactile resources and Hisense displays presenting calming visual content. Hisense’s advanced screen technology delivers clear, balanced visuals designed to support relaxation and sensory regulation.

Stadiums will feature sensory rooms within the venue or in the Stadium Fan Experience area as part of the expanded stadium footprint. In eight stadiums, both options will be available to fans, and fans will have access to a space in every stadium during every minute of the game itself. These rooms extend access to calming spaces throughout the venue, recognizing that sensory needs can arise at any moment during the event experience.

  • Ticket Access: In partnership with KultureCity, Hisense will provide complimentary match tickets in each Host City to families with sensory needs who may otherwise be unable to attend.

Creating a More Inclusive Tournament

In addition, the sensory rooms complement FIFA’s broader accessibility efforts, including sensory bags and trained venue staff to support fans with diverse needs.

The FIFA World Cup 2026™ will feature 104 matches across 16 cities over 39 days. For the first time in tournament history, every host stadium will include dedicated sensory-inclusive and accessible spaces — marking an important evolution in how global sporting events serve diverse audiences.

For more information about ticket applications through KultureCity, visit Hisense × KultureCity at FIFA World Cup 2026™ – KultureCity .

Leave a comment »

« Older Entries
Newer Entries »