Category: InfoSec

New Ghost Phishing Wave Is Breaking Traditional Email Security

Source: The Hacker News

Author: info@thehackernews.com (The Hacker News)

URL: https://thehackernews.com/2026/07/new-ghost-phishing-wave-is-breaking.html

ONE SENTENCE SUMMARY:

EvilTokens uses AES-GCM “ghost phishing” and Microsoft device-code flow to bypass URL checks, requiring browser-level sandbox visibility.

MAIN POINTS:

  1. Recent EvilTokens campaigns target US and European businesses with hidden “ghost phishing.”
  2. Malicious pages appear benign until decrypted and rendered inside the victim’s browser DOM.
  3. Attack leverages Microsoft Device Code Phishing to gain Microsoft 365 access without stealing passwords.
  4. AES-GCM encrypted HTML hides phishing content from static scanners and network inspection.
  5. Visibility gaps increase exposure time, delaying containment of Microsoft 365 account takeover.
  6. Compromised accounts risk unauthorized access to email, files, and cloud services.
  7. ANY.RUN sandbox revealed decrypted DOM behavior, Fetch/XHR activity, and device-code endpoints.
  8. In-browser inspection provides DOM snapshots, HTTP requests, URLs, and detection signatures.
  9. Extracted indicators include domains, endpoints, hashes, and infrastructure for threat hunting.
  10. Auto-generated reports speed Tier 1-to-Tier 2 handoffs and reduce duplicated investigation work.

TAKEAWAYS:

  1. Relying on email/URL “clean” results is insufficient against encrypted, browser-decrypted phishing.
  2. Device-code OAuth abuse enables stealthy account takeover with legitimate Microsoft login steps.
  3. Sectors with high phishing exposure face amplified risk from single Microsoft 365 credential compromise.
  4. Sandbox tooling must include in-browser data inspection to surface DOM changes post-decryption.
  5. Faster evidence-rich SOC workflows reduce incident costs and shorten the attacker’s dwell time.

Finding the “Goldilocks” Zone: A Practical Approach to Alert Triage

Source: Black Hills Information Security, Inc.

Author: BHIS

URL: https://www.blackhillsinfosec.com/the-goldilocks-zone/

ONE SENTENCE SUMMARY:

Effective incident triage balances urgency and efficiency by prioritizing severities, baselines, attacker objectives, detection intent, and contextual questions quickly consistently.

MAIN POINTS:

  1. Triage demands rapid assessment and clear classification to drive correct response decisions.
  2. Alert overload makes misclassification likely, so time management is critical.
  3. Low-severity findings usually provide poor investigative return and can often be ignored.
  4. Medium-severity alerts should be deferred until higher-priority signals shape investigation direction.
  5. High and Critical alerts typically reveal the core incident narrative and next steps.
  6. Baseline comparison quickly distinguishes normal behavior from true anomalies.
  7. Widespread “anomalies” across hosts may indicate expected operations or a broader problem.
  8. Evaluating attacker “actions on objective” highlights activity that advances exfiltration or lateral movement.
  9. Lack of meaningful progress toward goals often indicates benign behavior or non-impactful noise.
  10. Detection-intent focus reduces rabbit holes by validating only the specific TTP a rule targets.

TAKEAWAYS:

  1. Prioritize investigation time toward High/Critical alerts before revisiting Medium and Low.
  2. Use environment baselines to classify events faster and avoid chasing routine behavior.
  3. Look for goal-driven sequences like movement, escalation, or data access to confirm threat intent.
  4. Align analysis with what the detection rule actually tested for to improve investigation efficiency.
  5. Apply a consistent question-driven checklist: priority, frequency, attacker benefit, and success criteria.

Phantom Squatting Uses AI-Hallucinated Domains for Phishing and Malware

Source: The Hacker News

Author: info@thehackernews.com (The Hacker News)

URL: https://thehackernews.com/2026/07/phantom-squatting-uses-ai-hallucinated.html

ONE SENTENCE SUMMARY:

LLM-hallucinated domains enable “phantom squatting,” where attackers register predicted fake links, bypass reputation controls, and phish users.

MAIN POINTS:

  1. Attackers buy nonexistent AI-invented domains, then host phishing pages to capture traffic.
  2. Unit 42 names this technique phantom squatting and confirms real-world exploitation.
  3. Trust in AI-provided links lets criminals succeed without emails, ads, or traditional lures.
  4. Study queried two models 685,339 times across 913 brands and multiple industries.
  5. Responses contained 2.1 million links, including 13,229 already known malicious.
  6. About 250,000 hallucinated domains were unregistered, creating a large pre-registration target set.
  7. New domains evade blocklists because reputation systems need time and observed abuse.
  8. Models generate consistent hallucinations across temperatures, making attacker predictions easier.
  9. Case one: predicted domain registered 23 days later, used Montana Empire kit stealing IDs and payments.
  10. Case two: predicted domain registered 51 days later, used for brand-clone and malicious Android app distribution.

TAKEAWAYS:

  1. Monitor and preemptively watch likely hallucinated domains because defenders can gain weeks of warning.
  2. Verify official domains independently before entering credentials or using links in code.
  3. Prevent AI agents from auto-opening or downloading content from model-generated URLs without validation.
  4. Assume model output is a draft requiring confirmation, not a reliable authority.
  5. Recognize the broader “model output becomes input” shift accelerating phishing-as-a-service and response timelines.

Citrix patches a new NetScaler flaw with echoes of CitrixBleed

Source: CyberScoop

Author: Greg Otto

URL: https://cyberscoop.com/citrix-netscaler-flaw-cve-2026-8451-citrixbleed/

ONE SENTENCE SUMMARY:

Citrix disclosed six high-severity NetScaler flaws, led by CitrixBleed-like SAML memory disclosure, requiring patches and one post-patch configuration change.

MAIN POINTS:

  1. Tuesday’s Citrix bulletin covers six NetScaler ADC/Gateway vulnerabilities, overall rated high severity.
  2. CVSS scores span 6.9–8.8, indicating multiple serious attack paths across subsystems.
  3. CVE-2026-8451 leaks memory via out-of-bounds reads in SAML request parsing.
  4. Exploitation vector involves NetScaler configured as a SAML identity provider for SSO deployments.
  5. WatchTowr found CVE-2026-8451 while reproducing earlier CVE-2026-3055 from March.
  6. Root cause aligns with CitrixBleed-class issues: malformed SAML triggers memory disclosure conditions.
  7. Two additional CVEs are memory overflows that can cause denial-of-service.
  8. An unauthenticated arbitrary file-read affects appliances exposing management on certain interfaces.
  9. Another flaw is a TCP timestamp handling memory overread impacting NetScaler network processing.
  10. HTTP/2 malformed-request DoS needs patching plus manual timeout parameter adjustment for full remediation.

TAKEAWAYS:

  1. Prioritize patching NetScaler immediately, especially SAML IdP configurations handling authentication endpoints.
  2. Assume memory-safety weaknesses persist across releases; harden exposure and monitor aggressively.
  3. Restrict management interface reachability to prevent unauthenticated file-read opportunities.
  4. Verify post-update configuration changes, not just software versions, to fully mitigate HTTP/2 DoS.
  5. Although not confirmed exploited yet, NetScaler’s KEV history suggests rapid weaponization risk.

Your First GRC Agent: A Red Teamer’s Walkthrough

Source: BleepingComputer

Author: Sponsored by Anecdotes

URL: https://www.bleepingcomputer.com/news/security/your-first-grc-agent-a-red-teamers-walkthrough/

ONE SENTENCE SUMMARY:

Agentic AI transforms GRC into continuous, auditable control monitoring using autonomous, contextual agents that act on triggers while preserving human judgment.

MAIN POINTS:

  1. Widespread “agentic” hype obscures real operational changes in modern GRC programs.
  2. Legacy automation accelerates busywork but still produces static, periodic compliance artifacts.
  3. Agents differ by autonomy, contextual awareness, and multi-step analyze-decide-act execution.
  4. Modern environments demand real-time governance: elastic cloud, fluid identity, ephemeral infrastructure, nonstop CI/CD.
  5. Deterministic controls and human policy choices should govern AI orchestration and summarization.
  6. Practitioner work shifts from evidence collecting to higher-value judgment and control management.
  7. Continuous compliance becomes feasible when agents evaluate control state on change-triggered events.
  8. Trust and provability become bottlenecks once monitoring effort becomes cheap and ubiquitous.
  9. Building agents involves choosing triggers, writing plain-English instructions, then deploying with logs.
  10. Defensibility requires observable execution logs, least privilege, and human approval for consequential decisions.

TAKEAWAYS:

  1. Prioritize event-driven triggers to detect drift immediately rather than waiting for assessment cycles.
  2. Demand end-to-end traceability: inputs, rules evaluated, decisions, actions, and touched evidence.
  3. Constrain agent permissions and require human sign-off for closing risks or declaring control effectiveness.
  4. Expect mistakes and use logs to correct instructions, reducing false positives systematically.
  5. Start with low-judgment, high-toil tasks (evidence gaps, audit extraction) to build trust first.

Dangling CNAMEs: The Critical DNS Misconfiguration Most Organizations Still Miss

Source: Cloud Security Alliance

Author: unknown

URL: https://cloudsecurityalliance.org/articles/dangling-cnames-the-critical-dns-misconfiguration-most-organizations-still-miss

ONE SENTENCE SUMMARY:

Researchers found attackers hijacking trusted university subdomains via dangling CNAMEs, highlighting the need for continuous DNS posture management.

MAIN POINTS:

  1. Campaign abused abandoned CNAME records across major universities’ trusted .edu subdomains.
  2. Takeovers enabled hosting pornography, scams, fake apps, and malware delivery infrastructure.
  3. Dangling CNAMEs persist when DNS points to decommissioned cloud resources.
  4. Attackers reclaim missing cloud resource names to inherit control of organization subdomains.
  5. Trust in educational domains boosted SEO rankings, increasing exposure to malicious content.
  6. Cloud-era asset churn outpaces manual, fragmented DNS management, creating DNS sprawl.
  7. Separate teams manage cloud resources and DNS, leaving orphaned records unnoticed.
  8. Threat actors reportedly automate discovery and monetization of takeover-prone DNS entries.
  9. Traditional security tools miss inactive DNS dependencies and orphaned external relationships.
  10. Business damage is reputational, impacting search trust, deliverability, compliance, and brand credibility.

TAKEAWAYS:

  1. Prioritize monitoring external DNS dependencies as part of the organization’s attack surface.
  2. Automate detection and remediation of dangling CNAMEs across clouds and third-party platforms.
  3. Integrate DNS validation into infrastructure provisioning and deprovisioning lifecycle workflows.
  4. Adopt DNS Posture Management to continuously find orphaned records and takeover opportunities.
  5. Treat subdomain reputation as a security asset attackers can weaponize at low cost.

5 Claude Agent Skills Risks Every CISO Should Know

Source: Cloud Security Alliance

Author: unknown

URL: https://www.akto.io/blog/claude-agent-risks-for-cisos

ONE SENTENCE SUMMARY:

Claude SKILL.md agent skills mirror npm supply-chain risks, with widespread vulnerabilities, weak governance, over-privilege, stealthy markdown payloads, and cross-platform propagation.

MAIN POINTS:

  1. Claude Skills extend agents quickly but significantly expand the enterprise attack surface.
  2. Multiple audits show 26.1%–36.82% of skills contain at least one security flaw.
  3. OWASP introduced Agentic Skills Top 10 (AST10) in March 2026 to classify risks.
  4. Ungoverned installations create skill sprawl without inventory, SOC visibility, or centralized control.
  5. Silent skill loading from local directories enables unmanaged execution on developer workstations.
  6. Open registries allow poisoning via impersonation, malware uploads, and absent provenance mechanisms.
  7. Upstream compromise can trigger credential theft or execution merely by cloning untrusted projects.
  8. Skills inherit full agent context, causing over-privileged access to tokens, secrets, and PII.
  9. Markdown instructions can covertly direct exfiltration, evading traditional code-signature scanners.
  10. Cross-platform SKILL.md portability enables rapid reindexing, unsafe updates, and inconsistent permission behaviors.

TAKEAWAYS:

  1. Building a complete skill inventory is prerequisite to least privilege, scanning, and incident response.
  2. Treat skill registries like hostile supply chains without signing, verification, and publisher trust controls.
  3. Enforce per-skill isolation and sandboxing to prevent “one approval, infinite permissions” failures.
  4. Detect instruction-based threats by analyzing natural-language behaviors, not just executable code patterns.
  5. Containment requires immutable version pinning and update governance across every platform indexing SKILL.md.

GRC is broken. FedRAMP 20x might fix it

Source: GRC is broken. FedRAMP 20x might fix it | CSO Online

Author: unknown

URL: https://www.csoonline.com/article/4188995/grc-is-broken-fedramp-20x-might-fix-it.html

ONE SENTENCE SUMMARY:

Traditional compliance often audits curated snapshots; FedRAMP 20x and GRC engineering push continuous, machine-readable telemetry to restore trust through transparency.

MAIN POINTS:

  1. Compliance can become theatre when scope and narratives are managed for passing.
  2. SOC 2 and ISO 27001 are point-in-time snapshots, not maturity guarantees.
  3. Sampling-based audits miss drift, bypasses, and operational shortcuts outside the evidence window.
  4. “Passing audits does not equal security” because real behavior can diverge from documented controls.
  5. FedRAMP 20x targets automation-first assurance with machine-readable evidence and continuous validation.
  6. APIs and telemetry enable auditors to query complete datasets instead of curated screenshots.
  7. Exposing every VM, drift event, and posture history shifts focus to continuous posture maintenance.
  8. Full SDLC visibility reveals bypassed approvals and hotfix patterns hidden by selected pull requests.
  9. Identity lifecycle assurance improves by showing complete JML histories, not sampled access reviews.
  10. Auditor focus moves from control artifacts to validating evidence pipeline completeness and data integrity.

TAKEAWAYS:

  1. Optimize assurance around meaningful risk reduction, not “pass the audit” metrics.
  2. Build continuous evidence pipelines using APIs, telemetry, and structured, machine-readable outputs.
  3. Accept messy operational truth as a driver for improvement rather than a reputational threat.
  4. Adopt iteration loops for controls like engineering: measure, refine, and continuously validate.
  5. Prepare for trust models where customers and assessors query live assurance layers, not PDFs.

Cisco Umbrella Virtual Appliance Privilege Escalation Vulnerability

Source: Cisco Security Advisory

Author: unknown

URL: https://sec.cloudapps.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-umbrella-priv-esc-F4wJB7AU?vs_f=Cisco%20Security%20Advisory%26vs_cat=Security%20Intelligence%26vs_type=RSS%26vs_p=Cisco%20Umbrella%20Virtual%20Appliance%20Privilege%20Escalation%20Vulnerability%26vs_k=1

ONE SENTENCE SUMMARY:

Cisco Umbrella Virtual Appliance vmadmin CLI flaw enables authenticated local vmadmin users to gain root via command validation weakness, fixed by updates.

MAIN POINTS:

  1. Vulnerability affects Cisco Umbrella Virtual Appliance vmadmin CLI.
  2. Attack requires authenticated, local access to the device.
  3. Threat actor must already possess vmadmin privileges.
  4. Root cause is insufficient validation of user-supplied commands.
  5. Exploitation involves running specific commands through the CLI.
  6. Successful attack results in privilege escalation to root.
  7. Cisco issued software updates to remediate the issue.
  8. No workaround options are available for mitigation.
  9. Advisory rates the Security Impact Rating as Medium.
  10. Assigned identifier is CVE-2026-20246.

TAKEAWAYS:

  1. Prioritize patching affected appliances because compensating controls are unavailable.
  2. Restrict local access and vmadmin account usage to reduce exploit likelihood.
  3. Monitor and audit vmadmin CLI activity for suspicious command execution.
  4. Treat medium-severity local escalation as high operational risk on shared systems.
  5. Use the Cisco advisory link as the authoritative source for fixed versions.

Microsoft confirms Office apps launch issues after June updates

Source: BleepingComputer

Author: Sergiu Gatlan

URL: https://www.bleepingcomputer.com/news/microsoft/microsoft-confirms-office-apps-launch-issues-after-june-updates/

ONE SENTENCE SUMMARY:

Microsoft is investigating a Windows-update-triggered OLE automation issue blocking third-party apps from launching Office or opening documents reliably.

MAIN POINTS:

  1. A new known issue blocks third-party apps from launching Microsoft Office applications.
  2. Systems affected are fully updated with Windows patches released on/after June 9, 2026.
  3. Impacted Office products include Word, Excel, PowerPoint, Access, and others.
  4. Root cause involves third-party software using OLE automation to control Office.
  5. Failures may occur silently, with Office or documents not opening and no error.
  6. Reported affected apps include CCH Engagement, Zotero, and Workpaper Manager.
  7. Some vertical software is impacted, including dental tools like Dentrix and Softdent.
  8. Microsoft recommends opening Office apps/documents directly as a temporary workaround.
  9. Enterprises can request an organization-wide workaround via Microsoft Support for Business.
  10. A fix is being developed and will ship in a future Windows update.

TAKEAWAYS:

  1. Validate business-critical integrations that rely on Office OLE automation after June 2026 updates.
  2. Implement interim user guidance to launch Office directly rather than via third-party tools.
  3. Engage Microsoft Support for Business if widespread workflows are disrupted.
  4. Monitor Windows update releases for the promised remediation and related advisories.
  5. Expect silent failures, so add testing/monitoring to catch broken document-opening workflows.

SearchLeak: How We Turned M365 Copilot Into a One-Click Data Exfiltration Weapon

Source: Varonis Blog

Author: Dolev Taler

URL: https://www.varonis.com/blog/searchleak

ONE SENTENCE SUMMARY:

SearchLeak chained P2P prompt injection, streaming HTML race, and Bing SSRF to silently exfiltrate Microsoft 365 Copilot Enterprise data via one click.

MAIN POINTS:

  1. Varonis found a three-stage chain weaponizing Copilot Enterprise Search for covert data theft.
  2. Parameter-to-Prompt Injection lets the URL q parameter become executable Copilot instructions.
  3. Enterprise Search’s focus on organizational content makes it ideal for harvesting sensitive business data.
  4. Attack requires only one click on a trusted microsoft.com link—no plugins or permissions.
  5. A streaming-phase HTML rendering race allows <img> requests before output sanitization.
  6. Post-processing <code> wrapping occurs too late; the browser already emitted the outbound request.
  7. CSP blocks direct attacker domains, but *.bing.com is allowlisted for images.
  8. Bing’s “search by image” endpoint server-fetches attacker URLs, acting as an SSRF proxy.
  9. Exfiltration is achieved by embedding stolen mailbox content into the image URL path.
  10. Microsoft patched as CVE-2026-42824, rated critical, highlighting AI-enabled chaining of classic bugs.

TAKEAWAYS:

  1. Treat AI query parameters as potential instruction channels, not merely search input.
  2. Apply sanitization during streaming/render time, not after model output completion.
  3. Audit CSP allowlists for endpoints that perform server-side URL fetches.
  4. Detect suspicious Copilot Search URLs containing encoded HTML tags or exfiltration prompts.
  5. User awareness matters: long encoded Microsoft links and unsolicited Copilot searches warrant reporting.

Palo Alto Warns of Active Exploitation of PAN-OS GlobalProtect VPN Flaw

Source: The Hacker News

Author: info@thehackernews.com (The Hacker News)

URL: https://thehackernews.com/2026/06/palo-alto-warns-of-active-exploitation.html

ONE SENTENCE SUMMARY:

Palo Alto Networks reports limited active exploitation of PAN-OS CVE-2026-0257, urging log hunting, IoC blocking, and prompt mitigation.

MAIN POINTS:

  1. Palo Alto Networks observed active exploitation targeting GlobalProtect portals for unauthorized access.
  2. CVE-2026-0257 is an authentication bypass in PAN-OS portal and gateway components.
  3. The flaw enables attackers to bypass controls and initiate VPN connections.
  4. In-the-wild exploitation has been limited, first seen on May 17, 2026.
  5. Attribution remains unknown for the observed exploitation attempts.
  6. No post-access activity or lateral movement has been identified so far.
  7. Only a small subset of probed devices established VPN sessions and gateway-connected events.
  8. Published IoCs include multiple suspicious IP addresses tied to the activity.
  9. Additional IoCs list hostnames and MAC addresses associated with potential exploitation.
  10. CISA added the CVE to KEV, mandating FCEB mitigation by June 1, 2026.

TAKEAWAYS:

  1. Prioritize patching or mitigation for CVE-2026-0257 due to confirmed exploitation.
  2. Search GlobalProtect logs for successful gateway-connected events indicating compromise.
  3. Hunt for PoC-linked client values like Windows 10 Pro 64-bit and empty domain fields.
  4. Block and monitor provided IPs, hostnames, and MAC addresses in security controls.
  5. Use KEV deadlines to drive rapid remediation timelines and compliance reporting.

June Patch Tuesday marks a ‘new normal’ with over 200 CVEs, 32 rated ‘critical’

Source: June Patch Tuesday marks a ‘new normal’ with over 200 CVEs, 32 rated ‘critical’ | CSO Online

Author: unknown

URL: https://www.csoonline.com/article/4183632/june-patch-tuesday-marks-a-new-normal-with-over-200-cves-32-rated-critical.html

ONE SENTENCE SUMMARY:

June Patch Tuesday delivered record vulnerability volumes, including Microsoft’s 200+ CVEs, critical SAP flaws, and Adobe enterprise patches amid AI-accelerated discovery.

MAIN POINTS:

  1. Microsoft released fixes for over 200 CVEs, including three publicly disclosed zero days.
  2. Thirty-two Microsoft patches are rated critical, with additional high-risk flaws needing urgent assessment.
  3. Microsoft warns monthly CVE counts will keep rising and expects more out-of-band updates.
  4. AI-assisted discovery is shrinking time between bug existence and detection, pressuring patch cycles.
  5. Previously hard-to-audit areas like hypervisor code and Kerberos are yielding more vulnerabilities.
  6. Exchange Server CVE-2026-42897 moved from workaround guidance to an active-exploit patch.
  7. Microsoft flagged 15 flaws as “more likely” to be exploited, including http.sys kernel RCE CVE-2026-47291.
  8. High-rated Hyper-V VM escape vulnerabilities demand attention in virtualized enterprise environments.
  9. SAP issued 15 patches, including four critical vulnerabilities across core enterprise products.
  10. Adobe patched 123 vulnerabilities, highlighted by CVSS 10 issues in Campaign Classic and critical ColdFusion bugs.

TAKEAWAYS:

  1. Shift from slow patch testing to risk-based prioritization aligned with exploitation likelihood.
  2. Automate patching pipelines to handle sustained, “new baseline” vulnerability volumes.
  3. Treat internet-facing Windows services using http.sys as urgent remediation targets.
  4. Prioritize SAP ABAP/NetWeaver criticals due to high impact and low/no-auth attack paths.
  5. Include Adobe enterprise platforms (Campaign Classic, ColdFusion, Reader) in rapid patch SLAs.

ServiceNow Data Breach: Gated Advisory Left Customers Unaware of Exploited Zero-Auth API

Source: Tech Times

Author: unknown

URL: https://www.techtimes.com/articles/318166/20260610/servicenow-data-breach-gated-advisory-left-customers-unaware-exploited-zero-auth-api.htm

ONE SENTENCE SUMMARY:

ServiceNow fixed an unauthenticated API data-query flaw after confirmed exploitation, drawing criticism for delayed, gated disclosure and urgent customer response actions.

MAIN POINTS:

  1. Attackers queried sensitive ServiceNow customer tables via an unauthenticated API endpoint.
  2. Misconfiguration set requires_authentication=false, bypassing identity and privilege checks entirely.
  3. Reported exploited path was /api/now/related_list_edit/create against instance tables.
  4. Suspicious requests often came from IP 51.159.98.241, about five per tenant.
  5. Confirmed activity occurred June 2–3, 2026; hosted instances patched June 5.
  6. ServiceNow published advisory June 9, but hid it behind support-portal authentication.
  7. Limited notification meant many customers lacked a trigger to start incident response.
  8. Potentially exposed data includes tickets, HR records, assets, incident reports, tokens, and secrets.
  9. Credential leakage can enable lateral movement into integrated enterprise systems without further exploits.
  10. This is ServiceNow’s third major auth-related issue in eight months; first with pre-patch data access.

TAKEAWAYS:

  1. Search transaction logs for /api/now/related_list_edit/create and IP 51.159.98.241 during June 2–5.
  2. Review all Scripted REST Resources for requires_authentication disabled, including custom endpoints.
  3. Rotate credentials embedded in tickets and integrations immediately, assuming potential exposure.
  4. Enable verbose REST/API logging to support future scoping and exfiltration assessment.
  5. Engage legal counsel early since “queried vs exfiltrated” affects GDPR, HIPAA, and SEC obligations.

NIST AI RMF: Where to Start with AI Governance

Source: Rivial Security Blog

Author: Randy Lindberg

URL: https://www.rivialsecurity.com/blog/nist-ai-rmf-where-to-start-with-ai-governance

ONE SENTENCE SUMMARY:

Start AI governance with NIST AI RMF Govern, integrate into cyber risk, add inventories/controls, use FS AI RMF, report quantified ROI.

MAIN POINTS:

  1. AI adoption is accelerating faster than prior technology waves, embedding into core operations.
  2. Governance must begin immediately because AI appears via tools, vendors, and silent updates.
  3. NIST AI RMF provides four functions: Govern, Map, Measure, Manage.
  4. Prioritize Govern by establishing AI policy, ownership, procurement, and AI-aware change management.
  5. Build and maintain an AI system inventory tied to approvals, evidence, monitoring, and reporting.
  6. Fold AI risk into existing cybersecurity risk program, avoiding parallel AI risk silos.
  7. Extend eight cyber-risk elements with AI-aware updates, including AI-specific KRIs and TEVV.
  8. Quantitative risk measurement (e.g., Monte Carlo) beats qualitative heat maps for decision-making.
  9. FS AI RMF adds concrete, auditable controls useful beyond financial services for implementation.
  10. Board reporting should use dollars, expected loss reduction, and ROI, not red-yellow-green visuals.

TAKEAWAYS:

  1. Publish a workable AI policy now, then iterate as AI capabilities rapidly change.
  2. Treat AI as a risk dimension on existing systems, keeping one risk register and methodology.
  3. Counter shadow AI by offering approved tools and a fast intake/approval path.
  4. Add “accuracy” to CIA impacts to capture drift, bias, and hallucinations (CIA+A).
  5. Use breach-per-record and downtime estimates as ranges to quantify AI risk financially.

VS Code Adds 2-Hour Extension Auto-Update Delay to Limit Supply Chain Attacks

Source: The Hacker News

Author: info@thehackernews.com (The Hacker News)

URL: https://thehackernews.com/2026/06/vs-code-adds-2-hour-extension-auto.html

ONE SENTENCE SUMMARY:

VS Code now delays most extension auto-updates by two hours to reduce exposure to newly published supply-chain attacks.

MAIN POINTS:

  1. Microsoft introduced a two-hour delay for automatic VS Code extension updates.
  2. The change aims to mitigate software supply chain threats from compromised releases.
  3. This protection activates when automatic extension updates are enabled.
  4. The feature is available beginning with Visual Studio Code version 1.123.
  5. Users can still manually install updates immediately via the “Update” button.
  6. Extension details show why an update is pending and when it will occur.
  7. Trusted publishers’ extensions bypass the delay and update immediately.
  8. RubyGems added an opt-in cooldown to Bundler 4.0.13 for delayed gem installs.
  9. Bun, npm, pnpm, and Yarn added minimum-release-age controls with specific settings.
  10. Minimum-age thresholds reduce the spread window before malicious packages are detected and removed.

TAKEAWAYS:

  1. Delayed updates are emerging as a standard defense across developer ecosystems.
  2. A short cooldown can meaningfully limit exposure to fast-moving malicious releases.
  3. Manual update options preserve developer flexibility despite automated delays.
  4. Trust-based exceptions prioritize speed for major publishers, but increase reliance on publisher integrity.
  5. Supply-chain risk is rising, making proactive installation and update gating increasingly important.

Cybersecurity Hygiene Reinforced by the 2026 Verizon DBIR

Source: Blog Feed – Center for Internet Security

Author: unknown

URL: https://www.cisecurity.org/insights/blog/cybersecurity-hygiene-reinforced-by-the-2026-verizon-dbir

ONE SENTENCE SUMMARY:

Verizon’s 2026 DBIR shows CIS Controls and Benchmarks improve cyber hygiene, reducing exposure and countering prevalent modern attack techniques effectively.

MAIN POINTS:

  1. DBIR links common breach patterns to foundational security hygiene gaps across organizations.
  2. CIS Controls provide prioritized, actionable safeguards aligned to real-world attack paths.
  3. CIS Benchmarks harden system configurations, shrinking misconfiguration-driven compromise opportunities.
  4. Mapping DBIR top threats to CIS safeguards helps focus limited resources on highest risks.
  5. Strong identity, access management, and MFA reduce credential theft and account takeover impact.
  6. Vulnerability and patch management limit exploitation windows used by ransomware and initial access brokers.
  7. Secure configuration baselines improve consistency across cloud, endpoints, servers, and network devices.
  8. Continuous monitoring and logging support faster detection and response to prevalent intrusion techniques.
  9. Backup, recovery, and resilience controls blunt ransomware business impact and downtime.
  10. Governance and measurement using CIS frameworks enable repeatable hygiene improvements over time.

TAKEAWAYS:

  1. Prioritizing CIS Controls is a practical way to address the most frequent DBIR attack patterns.
  2. Implementing CIS Benchmarks reduces preventable breaches caused by insecure default configurations.
  3. Aligning security programs to evidence-based reports improves decision-making and investment justification.
  4. Standardized baselines and continuous verification are essential for sustaining cybersecurity hygiene.
  5. Combining prevention, detection, and recovery controls provides better defense against modern, multi-stage attacks.

Hackers Are After the Gaps in Your Vulnerability Program: Here’s Their Playbook

Source: BleepingComputer

Author: Sponsored by Flare

URL: https://www.bleepingcomputer.com/news/security/hackers-are-after-the-gaps-in-your-vulnerability-program-heres-their-playbook/

ONE SENTENCE SUMMARY:

Underground tutorial by “Hercules” teaches novices to find, validate, and monetize vulnerabilities, spreading widely and challenging defenders’ patching programs worldwide.

MAIN POINTS:

  1. Forum post presents a simple end-to-end workflow: scan, assess, exploit, monetize.
  2. Author emphasizes tracking newly disclosed high-impact flaws like RCE, auth bypass, ATO.
  3. Guidance includes locating exposed systems and verifying vulnerability status at scale.
  4. Nuclei framework and community templates are promoted for fast, automated discovery.
  5. Tutorial explicitly separates “legal” disclosure paths from “illegal” exploitation choices.
  6. Plain-language tone lowers barriers, framing hacking as learnable through practice, not theory.
  7. Responses show beginners seeking mentorship, private contact, and applied guidance.
  8. Method’s popularity led to reposts and discussion across four additional underground forums.
  9. Monetization options include paid disclosure, underground sales, or direct exploitation for access.
  10. Discussion highlights defender patching delays and the persistent risk of legacy vulnerabilities.

TAKEAWAYS:

  1. Simplified, repeatable playbooks can scale cybercrime more than novel techniques.
  2. Rapid patching and exposure management matter most for reachable critical vulnerabilities.
  3. Old, unmaintained platforms remain profitable targets because novices can exploit known CVEs.
  4. Well-designed paid disclosure programs can shift incentives toward reporting over exploitation.
  5. Threat intelligence should monitor tutorial传播 and recruitment behaviors, not just IOCs.

7 tabletop exercise mistakes that sabotage incident response

Source: 7 tabletop exercise mistakes that sabotage incident response | CSO Online

Author: unknown

URL: https://www.csoonline.com/article/4179644/7-tabletop-exercise-mistakes-that-sabotage-incident-response.html

ONE SENTENCE SUMMARY:

Effective cyber tabletop exercises require clear objectives, realistic ambiguity, business-specific detail, right stakeholders, and testing interdependent decisions—not compliance theater alone.

MAIN POINTS:

  1. Running tabletops without measurable objectives rewards improvisation and obscures plan effectiveness.
  2. Generic ransomware scripts cause exercises to drift into discussion rather than readiness testing.
  3. Practicing only familiar incidents leaves teams unprepared for ambiguous, conflicting real-world signals.
  4. Introducing incomplete information forces decision-making under uncertainty, mirroring actual breach conditions.
  5. Scenarios must reflect your environment, priorities, past incidents, and industry threats.
  6. Missing stakeholders—legal, communications, HR, operations, executives—creates gaps in ownership and escalation.
  7. Capturing stalls, unclear decision rights, and absent voices should drive iterative improvements.
  8. Technical implausibility erodes buy-in; attack chains must logically connect to real architecture.
  9. Overly theoretical prompts high-level answers, hiding tooling gaps, authority limits, and communication breakdowns.
  10. Stress-test handoffs and dependencies across teams and vendors using risks from the organization’s register.

TAKEAWAYS:

  1. Define success criteria around escalation, notification, decision rights, and recovery prioritization before the exercise.
  2. Design ambiguity deliberately so participants practice choosing actions with partial, conflicting data.
  3. Tailor scenarios to business realities and include all real incident stakeholders.
  4. Ground narratives in technically accurate details that match systems, logs, and likely attacker paths.
  5. Measure outcomes, document friction points, and update plans and future tabletops to close gaps.

Attackers exploit Palo Alto GlobalProtect flaw days after disclosure

Source: Attackers exploit Palo Alto GlobalProtect flaw days after disclosure | CSO Online

Author: unknown

URL: https://www.csoonline.com/article/4179847/attackers-exploit-palo-alto-globalprotect-flaw-days-after-disclosure.html

ONE SENTENCE SUMMARY:

Attackers exploit CVE-2026-0257 in Palo Alto GlobalProtect, bypassing authentication via forged cookies, accelerating patch urgency and zero-trust scrutiny.

MAIN POINTS:

  1. Active in-the-wild exploitation followed Palo Alto’s initial medium-severity disclosure within days.
  2. Rapid7 observed successful VPN access across customers, without confirmed lateral movement.
  3. CVE-2026-0257 impacts GlobalProtect remote-access VPN on PAN-OS devices.
  4. Exploitation reportedly began May 17, shortly after fixes and mitigations were published.
  5. Palo Alto raised CVSS to 7.8, marked “attacked,” and set highest urgency.
  6. Vulnerability enables credential-less authentication bypass by forging a trusted cookie.
  7. Sessions appear legitimate, complicating detection compared with typical intrusion methods.
  8. Root cause: decrypted cookie contents trusted without signature verification.
  9. Exposure requires specific configuration: override cookies enabled and shared certificate usage.
  10. CISA added it to KEV, ordering rapid remediation for federal agencies.

TAKEAWAYS:

  1. Treat auth-bypass flaws on remote-access gateways as critical, regardless of base scoring.
  2. Audit GlobalProtect configurations for authentication override cookies and certificate reuse.
  3. Patch immediately and apply mitigations; exploitation can start days after disclosure.
  4. Strengthen monitoring for suspicious “legitimate” VPN sessions that may be forged.
  5. Improve asset visibility and configuration governance to reduce edge-device exposure during zero-trust transitions.

Top 6 Claude Security Risks to Watch as AI Becomes Your Employees’ Operating System

Source: Cloud Security Alliance

Author: unknown

URL: https://www.akto.io/blog/claude-security-risks

ONE SENTENCE SUMMARY:

Claude’s expanding privileges create shadow AI, connector, skills, and code risks requiring comprehensive discovery, governance, IAM, SDLC controls, and monitoring.

MAIN POINTS:

  1. Unapproved Claude usage exposes proprietary, financial, and legal data without organizational visibility or guardrails.
  2. Missing SSO and acceptable-use policies prevents understanding data flows and regulatory compliance status.
  3. Claude Projects act as unmanaged repositories for sensitive documents, access sharing, and connectors.
  4. Organizations often cannot identify uploaded files, project access holders, or active connector activity.
  5. MCP connectors expand attack surface by enabling direct access to Slack, GitHub, Drive, Jira, Notion.
  6. OAuth scopes and authentication boundaries are frequently over-permissioned by users for convenience.
  7. Cowork introduces autonomous AI actions, complicating accountability, policy enforcement, and auditing requirements.
  8. Claude Code skills create supply-chain risks; plain-English prompts can drive data exfiltration.
  9. Studies found high vulnerability rates in Claude-generated code, increasing production security defects.
  10. Platform flaws in Claude Code enable malicious repositories to trigger command execution and key compromise.

TAKEAWAYS:

  1. Perform enterprise-wide asset discovery to inventory Claude usage across web, desktop, Code, and Cowork.
  2. Treat Projects as persistent data stores and enforce DLP with classification and real-time monitoring.
  3. Govern MCP/connector enablement with security review, least privilege, and token-usage visibility.
  4. Apply secure SDLC gates to AI-generated code, skills, extensions, and autonomous workflows.
  5. Build continuous audit trails for AI activity, access patterns, and sensitive-data exposure across all surfaces.

Grading on a curve: How to assess a pentest

Source: The Red Canary Blog: Information Security Insights

Author: Brian Donohue

URL: https://redcanary.com/blog/testing-and-validation/pentesting/

ONE SENTENCE SUMMARY:

Effective defense disrupts multi-stage attack chains by prioritizing high-fidelity, intent-rich behaviors, not exhaustive detection of every atomic action.

MAIN POINTS:

  1. Breaches result from campaigns of sequential actions, not single attacker successes.
  2. Detecting any critical step can hinder adversaries, evict threats, and prevent incidents.
  3. Depth and redundancy help, but complete coverage of all behaviors isn’t required.
  4. Testing is often misused as an exhaustive scorecard demanding alerts for every action.
  5. Real threats are adaptive, persistent campaigns; emulations are usually partial and constrained.
  6. Defensive focus should be “detect to disrupt” by breaking the attack chain early.
  7. Early or mid-chain detection can outperform noisy reconnaissance detection in outcomes.
  8. Isolated atomic events lack context; patterns reveal malicious intent and progression.
  9. High-fidelity TTPs like LSASS dumping and persistence provide reliable intervention points.
  10. Over-alerting to catch everything increases false positives and reduces analyst effectiveness.

TAKEAWAYS:

  1. Measure success by stopping attacker objectives, not by maximizing alert counts.
  2. Prioritize chokepoints and intent-rich techniques that reliably indicate malicious progression.
  3. Treat pentests and red teams as validation inputs, not comprehensive real-threat replicas.
  4. Use contextual correlation to distinguish benign activity from adversary behavior patterns.
  5. Expand coverage thoughtfully to scale, avoiding alert floodgates that bury true threats.

The Alert Firehose Finally Meets Its Match

Source: The Hacker News

Author: info@thehackernews.com (The Hacker News)

URL: https://thehackernews.com/2026/05/the-alert-firehose-finally-meets-its.html

ONE SENTENCE SUMMARY:

Agentic AI-enhanced NDR converts high-volume network telemetry into correlated, prioritized detections, reducing false positives and accelerating SOC response.

MAIN POINTS:

  1. Legacy NDR earned a “noisy” reputation from raw visibility and SIEM-overloading alerts.
  2. Manual tuning demands during deployment often determined whether NDR became actionable or overwhelming.
  3. Agentic AI autonomously gathers context, triages alerts, and correlates evidence across events.
  4. High data volume shifts from burden to advantage when AI can analyze thousands of signals.
  5. Correlation links low-severity activities into threat narratives humans typically miss.
  6. AI produces prioritized detections with supporting network evidence for immediate analyst context.
  7. Automation reduces dependency on extensive manual tuning by learning detection improvements.
  8. Example scenario: hundreds of anomalies collapse into four actionable detections after AI triage.
  9. Effective operations still require baselining, ongoing tuning, and SOC workflow integration.
  10. Data quality drives AI security outcomes more than model choice, improving accuracy and findings.

TAKEAWAYS:

  1. Reputational “noise” issues reflect earlier operational realities, not modern AI-enabled NDR performance.
  2. Correlation at scale is the differentiator that turns telemetry into actionable threat stories.
  3. Maintaining an updated baseline prevents environmental changes from becoming false-positive generators.
  4. Feeding other SOC tools with pre-correlated NDR outputs reduces downstream alert fatigue.
  5. Success depends on deployment discipline plus high-fidelity network data powering AI-driven analysis.

FBI warns of Kali Oauth stealers

Source: FBI warns of Kali Oauth stealers | CSO Online

Author: unknown

URL: https://www.csoonline.com/article/4176464/fbi-warns-of-kali-oauth-stealers.html

ONE SENTENCE SUMMARY:

FBI warns Kali365 phishing steals Microsoft 365 OAuth tokens, bypasses MFA via device authorization, urging conditional access blocks and transfer restrictions.

MAIN POINTS:

  1. FBI alerted organizations about a new Kali365-enabled phishing wave targeting Microsoft 365 accounts.
  2. Kali365 captures OAuth access tokens rather than stealing usernames or passwords.
  3. Bypassing multi-factor authentication occurs because valid tokens authenticate without credential interception.
  4. Attackers impersonate trusted cloud document-sharing services in convincing phishing emails.
  5. Victims are instructed to enter a specific code on a legitimate Microsoft website.
  6. Entered code authorizes the attacker’s device to access the victim’s Microsoft account.
  7. Mitigation includes conditional access policies blocking device code flow for most users.
  8. Exceptions should be narrowly granted only for essential business processes needing code flow.
  9. Blocking authentication transfer policies prevents rights handoff from corporate PCs to mobile devices.
  10. World Economic Forum data shows phishing is CEOs’ top concern and growing across organizations.

TAKEAWAYS:

  1. Token-based phishing can defeat MFA without ever capturing user credentials.
  2. Legitimate login pages don’t guarantee safety when attackers abuse device authorization workflows.
  3. Conditional access controls are central to reducing exposure to device code phishing.
  4. Preventing authentication transfers limits attackers’ ability to persist across devices.
  5. Rising phishing volume makes rapid policy hardening and user awareness critical.

Critical vulnerability in Cisco Secure Workload rated at maximum severity

Source: Critical vulnerability in Cisco Secure Workload rated at maximum severity | CSO Online

Author: unknown

URL: https://www.csoonline.com/article/4175913/critical-vulnerability-in-cisco-secure-workload-rated-at-maximum-severity.html

ONE SENTENCE SUMMARY:

Cisco Secure Workload on-prem has a CVSS 10 auth-bypass REST API flaw granting site-admin control, requiring immediate patching.

MAIN POINTS:

  1. Vulnerability enables attackers to gain site admin privileges and compromise endpoints.
  2. Cisco Secure Workload controls zero trust, micro-segmentation, and network visibility across enterprises.
  3. Threat actors likely will scan aggressively for exposed, unpatched internal API endpoints.
  4. Site-admin access could modify or dismantle security policies, opening previously restricted pathways.
  5. Multi-tenant deployments face cross-tenant impact, expanding potential exposure across business units or customers.
  6. CVE-2026-20223 has CVSS 10.0, allowing unauthenticated remote authentication bypass.
  7. Crafted HTTP requests to internal REST APIs instantly confer site admin privileges.
  8. Root cause is insufficient validation and authentication on REST API endpoint access.
  9. No workarounds exist; only software updates remediate the issue.
  10. SaaS is already patched, while on-prem customers must upgrade to fixed releases.

TAKEAWAYS:

  1. Prioritize emergency patching for on-prem Secure Workload as if responding to an active incident.
  2. Upgrade targets: 4.0→4.0.3.17, 3.10→3.10.8.3, 3.9 and earlier→migrate forward.
  3. Focus assessment on internal REST API exposure rather than the web management interface.
  4. Treat multi-tenant environments as higher-risk due to potential cross-tenant “blast radius.”
  5. Verify patch status promptly despite no known exploitation reported at disclosure time.