Search the archive

FULCRUMSEC - A LOOK AT THEIR TRADECRAFT

FulcrumSec, a financially motivated data extortion gang behind breaches at Manchester Airports Group, Arup, and Novo Nordisk, told Lab539 it runs up to 250 servers scanning client-side JavaScript and fresh bugs like React2Shell, then bulk-tests harvested credentials before picking victims. From there, it chains pivots through GitHub repos, AWS, Azure, and Secrets Manager without touching an endpoint, with a single Arup PAT reportedly opening about 10,000 private repositories. Defenders should baseline normal identity behavior and flag tokens that suddenly clone thousands of repos or enumerate unrelated secrets from unfamiliar infrastructure, since this activity relies on valid credentials rather than malware.

BELTDOWN2: ESCAPING THE CURSOR CLI SANDBOX

A malicious workspace can place a core.fsmonitor hook in .git/config. Cursor CLI ran internal Git commands outside its macOS Seatbelt sandbox. A read-only prompt triggered git ls-files, which executed the hook. The proof payload wrote to $HOME, while sandboxed shell commands could not. Anysphere acknowledged the report and shipped Cursor CLI 2026.08.04-aaa8809, and the fix injects GIT_CONFIG environment settings into every Git spawn, disabling core.fsmonitor, custom hooks, and repository attributes.

UNMASKING CLOUD IDENTITIES: FROM BEHAVIORAL CLUSTERING TO AUTOMATED DETECTION

Unit 42 clustered the CloudTrail activity of over 40,000 identities across 125 AWS environments and found they fall into consistent functional roles like console admins, CI/CD runners, backup agents, and security tools, regardless of naming or assigned IAM policies. That behavioral baseline matters because attackers masquerade behind benign labels and existing permission profiles, so a backup service suddenly running admin actions stands out even when its permissions allow it. The approach distills into a lightweight classifier that runs as standard SQL, letting teams tag identity roles at scale without a full ML pipeline.

running Go-based SSH. Integrity matters. Read more →

OPENHUNTERAI

OpenHunterAI is a local AI red-team workspace that runs attacker-style assessments against web apps, APIs, and LLM applications you own, combining browser inspection, recon, ZAP, and a Nuclei adapter under one scan plan. Its differentiator is scope control, with domain verification, deterministic plans, and human approval gates before any sensitive validation runs. It also ships as a skill for Claude Code, Codex, and Gemini CLI, so coding agents can prepare scoped assessments that wait for human reviewers to sign off.

RF-SWIFT

A powerful multi-platform RF toolbox that deploys specialized radio, hardware, and other security tools in seconds on Linux, Windows, and macOS - supporting x86_64, ARM64 (Raspberry Pi and Apple Silicon), and RISC-V architectures without disrupting your primary OS.

AIUC (PRODUCT LAUNCH)

AIUC audits and certifies deployed agents against its AIUC-1 standard. Its service runs roughly 5,000 tests covering jailbreaks, hallucinations, and data leaks. Customers receive a 100-page report, with humans validating final audit results.

TWO DGX SPARKS, 33 LOCAL MODELS, ONE QUESTION: CAN LOCAL AI ACTUALLY RED TEAM?

Roei Sherman evaluated 33 local models on a two-node DGX Spark cluster for red teaming capabilities. The evaluation consisted of a security capability that tested whether the models can reason by testing them with a three-phase red team scenario, a red team benchmark to test the willingness and accuracy of model responses, and a coding benchmark consisting of Aider Polyglot plus two custom suites. Sherman found that MoE models consistently performed above their model class across all three benchmarks, with Qwen3.6-35B-A3B (FP8) winning the security capability benchmark, Qwen3-Coder-Next (FP8) winning the coding benchmark, and more mixed results on the red team benchmark.

MYTHIC C2 ACTIVITY AT INTERNET SCALE

Mythic, an open-source C2 framework that lets operators mix agents, transport profiles, and a shared web UI, is popular with red teams but has also turned up in real intrusions, according to new Censys research. Censys tracked 131 internet-exposed Mythic hosts, and 115 of them still carried the default O="Mythic" or O="Mythic C2" certificate. One cluster Censys flagged paired Mythic with custom Rust implants, Discord-based C2, and a domain disguised as Windows telemetry traffic.

controls fail and how to close the gap.

MACOS 27 GOLDEN GATE, MACOS TAHOE 26.7, AND MACOS SEQUOIA 15.8 FIX 200+ VULNERABILITIES

Apple patched more than 200 macOS vulnerabilities across Golden Gate 27, Tahoe 26.7, and Sequoia 15.8, including kernel code execution bugs in AVEVideoEncoder and UDF, as well as remote code execution paths via Bluetooth, CUPS, and WebDAV.

240,000 HIT BY DATA BREACH AT JAPAN'S DIGITAL AGENCY

Japan's Digital Agency found that attackers exploited a publicly disclosed VPN flaw.

AUTOMATED SECURITY REVIEW FOR PLUGIN RELEASES

WordPress.org scans each plugin release during a six-hour cooldown before update distribution.

Register →

View the agenda →

DREAMFORCE 2026: SLACK'S NEWEST INNOVATIONS TURN AI INTO A TRUE TEAMMATE

Slack is expanding far beyond chat with Slackbot in channels, full-screen AI workspaces, live Salesforce-powered Surfaces, and Slack Code, where developers and AI agents can build and review code collaboratively. Slack is becoming a control surface for more business systems and agents, raising new questions around app governance, permissions, licensing, and what employees should be allowed to connect.

WINDOWS 11 KB5129195 IS OUT AFTER MICROSOFT CONFIRMS MAJOR ISSUES WITH THE SEPTEMBER 2026 UPDATE, BUT IT WON'T FIX AMD GPU ERRORS

Microsoft has released an out-of-band Windows 11 update after September's Patch Tuesday caused problems with Remote Desktop, audio, virtualized apps, and other components. KB5129195 fixes several of the issues, but reported Explorer crashes, File History failures, some USB audio problems, and AMD GPU instability remain worth watching before accelerating enterprise deployment.

AI WAS SUPPOSED TO MAKE SOFTWARE CHEAPER. SO WHY IS IT MORE EXPENSIVE NOW?

Despite dramatic improvements in developer productivity and falling unit costs for AI intelligence, software is becoming more expensive due to rising software scope and Jevons paradox driving up consumption. Software has historically been priced on the value delivered to the buyer rather than production labor costs. Consequently, efficiency gains are absorbed by increased inference usage and higher feature expectations rather than lower bills.

WHEN YOUR CRM STARTS BUILDING ITSELF: THE CIO'S ROLE IN THE AGENTIC SAAS ERA

AI is turning business users into application and agent builders faster than traditional IT processes can keep up. Instead of trying to centrally build everything, CIOs increasingly need to own the foundations - identity, data architecture, observability, integration standards, and agent governance - while letting business teams build safely on top of them.

DREAMFORCE OPENS WITH AI, BUT QUESTIONS FOR CIOS REMAIN

Salesforce's Dreamforce announcements point toward enterprise software being accessed increasingly through agents, MCP, and conversational interfaces rather than traditional application screens. The unresolved IT question is economic: as software moves away from predictable seat-based licensing toward agent usage, MCP calls, and consumption pricing, CIOs may gain flexibility while losing some cost predictability.

MEET ADAPT, WESTPAC'S AZURE-BASED ENTERPRISE DATA PLATFORM

Westpac's Adapt is a new enterprise data platform built on Microsoft Azure that consolidates data from 285 source systems into a single governed environment. The platform required migrating over 1PB of data and 14,000 data pipelines. Additionally, Westpac utilizes a Snowflake-powered intelligence layer to run advanced AI models on data at rest without moving it.

INTRODUCING BUILDER CENTRAL: THE NO CODE AI WORKSPACE FOR SALESFORCE

Salesforce Builder Central is a no-code AI workspace designed to let admins and business users build and configure apps and agents using plain language. Powered by Headless Toolkit, the workspace natively reads company metadata and Data 360 context while providing built-in trust, security, and automatic safety checks. The beta version is scheduled to launch the week of September 21.

BUILD WITH CONFIDENCE: APP GOVERNANCE, ACCESS, AND ACCOUNTABILITY AT HUBSPOT

HubSpot is adding centralized controls for which apps and AI connectors can be installed, who can install them, what data they can access, and who owns each integration. Particularly useful for IT: admins can require approval, use more granular OAuth scopes, restrict MCP access to company-provisioned accounts, see app activity logs, and get warned when deactivating someone who owns an integration.

ZOOM UNVEILS AI-POWERED REVENUE OS TO UNIFY BUYER INTELLIGENCE, CUSTOMER CONVERSATIONS, AND REVENUE EXECUTION

Zoom is expanding its platform with Common Room integration and new Revenue Accelerator capabilities for engagement and forecasting. It is another example of collaboration vendors broadening into adjacent enterprise workflows, potentially giving IT teams more opportunities to consolidate tools, but also more overlap to evaluate across CRM, meeting intelligence, and sales platforms.

OBSIDIAN VAULT TO ENTERPRISE COMPANY BRAIN: WHERE AGENTS WRITE, BRANCH, AND MERGE

Scaling knowledge management from a personal Obsidian graph to an enterprise-grade "company brain" requires AI agents to manage knowledge without corrupting it through hallucinations or conflicting updates. Versioned branches, object storage, and a typed graph schema can keep changes controlled, with human review governing merges.

AI CLOUD BILLS ARE FORCING TECHNOLOGY LEADERS TO RETHINK AWS, AZURE AND GCP COMMITMENTS

AI workloads are disrupting traditional cloud commitments as GPU availability, data movement, egress fees, and unpredictable inference demand make long-term AWS, Azure, and GCP agreements harder to forecast. CIOs are increasingly separating AI infrastructure economics from general cloud spend, negotiating more flexible capacity terms, and deciding which high-cost workloads need to remain portable across providers

MORE JFROG ARTIFACTORY BUGS ARE UNDER ATTACK, AND ALL THREE HAVE PATCHES

Researchers at Wiz report active exploitation of three JFrog Artifactory vulnerabilities, including an authentication-bypass bug and two flaws being chained together for privilege escalation.

SCALING CLAUDE WITH COST CONTROLS (45 MINUTE VIDEO)

Anthropic walks admins through new Claude Enterprise controls for model defaults, per-user spend visibility, usage reporting, and managing consumption-based AI costs without a dedicated FinOps operation.

MICROSOFT TEAMS: EXTERNAL PRESENTERS CAN MANAGE EVENT PRODUCTION CONTROLS

Teams organizers can give trusted B2B and guest presenters production controls for events, reducing the need to provision external event-production partners as internal users.

Get (virtual) front-row access to

* What connected advertising is and how it reshapes marketing work * Linking AI to business value, not just productivity gains * StackAdapt's latest AI innovations, including Ivy Studio Keynote speakers include StackAdapt's Co-Founder and CEO Vitaly Pecherskiy, OpenAI's VP of Monetization Benji Shomair, AI/ML Computer Scientist & Educator Harper Carroll and more.

Register now to reserve your spot

COMPANIES ARE HIRING EDITORS TO RUN THOUGHT LEADERSHIP AND MEASURING THEM ON PIPELINE

Companies are hiring senior editors to run thought leadership and tying the function to revenue. Workato, for example, measures its program on win rates, deal velocity, and average sales price instead of content output. Research shows thought leadership can influence buying decisions and reach “hidden buyers” who rarely interact with sales. Recurring research and publishing can also build category authority over time. Owned media is becoming a permanent function with editorial leadership and a pipeline-focused scorecard.

GEN Z ROMANTICISED THE LIBRARY INTO RELEVANCE AGAIN

Gen Z is driving a library revival through viral TikTok and Instagram content. Creators frame libraries as aesthetic, free places to study, work, and socialize. 67% report visiting public libraries, including many who are not active readers. The trend also ties into growing interest in physical books and digital detox. For libraries, social content can turn cultural relevance into real-world visits and stronger evidence for continued public funding.

I AUDITED 4 WEEKS OF MY LINKEDIN POSTS. HERE'S WHAT GREW MY FOLLOWING

A 4-week LinkedIn audit found that conversation-starting posts, relatable humor, and provocative takes drove the most follower growth. Shorter copy also performed better than long paragraphs. Memes were fast to produce and often earned strong engagement, but they didn't build followers or trust as effectively as video. The best mix is repeatable, high-engagement formats like memes alongside videos and posts that spark debate or make the audience feel seen.

DON'T USE QUESTIONS IN YOUR TITLES

Use statement-based headlines instead of questions to increase engagement with marketing content. A Stanford study found that question titles generated 71.6% fewer comments and 12.6% fewer clicks across experiments and analyses of more than 22,000 news articles, 3 million academic articles, and 53,000 Reddit posts. Questions may perform worse because they signal uncertainty and provide less immediate information to readers. Test statement formats such as “X compared to Y” against question-based titles, then check that title changes don't hurt SEO.

STOP TRYING TO GO VIRAL. YOU NEED THE RIGHT 100 PEOPLE TO KNOW WHO YOU ARE FIRST

Personal branding isn't a numbers game, it's a positioning game. Being known for one specific thing, not several, is what makes a name and a topic become inseparable in someone else's mind. The first time an idea gets posted, it's scrolled past. The fifth time, it starts to register, and by the twentieth repetition a name becomes the person others associate with that topic. That kind of association compounds into leverage long before any follower count does.

across the surfaces that matter. REGISTER NOW

AI OVERVIEWS AND PAID SEARCH: WHEN TO FIGHT, INFLUENCE, OR GENERATE DEMAND

AI Overviews make it harder to use one bidding strategy across every search query, so marketers should classify terms by intent and AI visibility before allocating budget. High-intent terms belong in the “Fight” bucket, where paying for absolute top placement can still generate direct sales. Informational terms belong in “Influence”, where SEO and PPC should work together to earn citations and brand mentions in AI answers. Zero-click opportunities belong in “Generate Demand”, where budget should move to YouTube, Demand Gen, and content that builds demand before the eventual commercial search. In one B2B account, 83.3% of queries triggering AI answer blocks were informational, showing why reallocating budget across these buckets can matter more than simply adjusting bids.

SALESFORCE AND NVIDIA'S NEW REASONING MODEL IS EVERYTHING THE AI LABS SHOULD FEAR

Koa answers a problem Salesforce couldn't solve without Nvidia, a reasoning model trained for sales and support work that never has to touch a customer's actual data. It was post-trained entirely on synthetic conversations, including simulated irate customers and negotiating sales reps, and its architecture is built to reach a token-efficient answer faster than a general-purpose frontier model would. Its own AI gateway can now route reasoning-heavy requests to Koa instead of paying frontier-lab rates for the same work.

DOES TV ACTUALLY SUCK NOW?

Streaming didn't shrink television's quality - it multiplied its supply, turning a manageable weekly lineup into a catalog large enough to produce real decision fatigue. An audience now half-watching on a second screen while scrolling through that catalog is bound to feel worse about whatever it lands on, regardless of how well made it actually is. The fix is a shift in viewing habits, not a demand on the industry to make better shows.

THE JOB MARKET HAS BEEN CONTRACTING FOR 31 MONTHS

White-collar hiring has contracted for 31 months, with professional and technical roles shrinking despite stable overall employment. The bigger change is repeated headcount deferrals, leaving existing teams to absorb work indefinitely. Instead of waiting for hiring to recover, identify what you're actually paid for beyond your job title. Turn those skills into something clients or smaller companies would pay for. Build a small version while still employed, then review your progress in 6 months.

WEIRD ECOMMERCE SEO TIP: GO TO THE SHOPS

The physical taxonomy of store aisle signage can inform ecommerce website structure.

HERE'S HOW & WHEN I RUN GEO-LIFT TESTS

A mental model for geo-lift testing on Meta ads.

HUMANIST AI IN PRACTICE: A PUBLIC CONSULTATION ON OUR CODE OF CONDUCT FOR MAI MODELS

Microsoft published a first draft of its AI training and deployment manual.