Security Leadership

An independent, attributed roundup of the major cybersecurity incidents of October 2026 — from the PHP2Shell critical RCEs and the ConnectWise Cloud Backup supply-chain compromise to breaches at ETS, Denmark’s population register, Zeabur, Railway, Burger King Japan and more. Each item carries a verification status (confirmed, claimed or rumoured) and a clearly-labelled subjective editorial severity, with a defensive takeaway.

PwC has confirmed a cybersecurity incident at a third-party vendor that hosts its employee learning records, after the Everest extortion group listed PwC and claimed ~590.69 GB of data. What PwC actually confirmed, what Everest claims (still unverified), and the third-party / LMS lessons for security leaders.

A practical guide to security metrics for 2026: what makes a metric good, how KPIs differ from KRIs and KCIs, leading versus lagging indicators, how to report differently to the board, the CISO and the SOC, the classic core set worth tracking, and the AI-era metrics most programs are still missing. No made-up industry averages — learn how to set your own targets instead.

A step-by-step guide to building a security metrics program that executives trust and the SOC actually uses: tie metrics to decisions, pick a small core set per audience, define each metric precisely, baseline before you set targets, automate collection, report differently to the board and the SOC, and review and retire vanity metrics. Includes a section on adding AI-era metrics. No fabricated statistics — a framework for setting your own.

On 21 October 2026 GitLab shipped a critical fix for CVE-2026-110264, a CVSS 10.0 unauthenticated remote code execution bug in the bundled regex compiler that affects self-managed CE/EE back to version 7.0 (2014). Instances that only applied the earlier 12 October regex fix are still exposed and must upgrade again. What it is, who is affected, how to check, and what to do now.

On 21 October 2026, malicious LiteLLM 1.107.0 artifacts hit PyPI, Docker Hub and GHCR. The payload ran at Python startup via a .pth file — no import needed — and exfiltrated cloud, AI, SSH and GPG credentials to a look-alike domain. What happened, the IOCs, and exactly what a security team should do now.

A methodology explainer for security leaders: how cyber-threat-intelligence analysts correlate leak-FORUM POST METADATA — not the leaked data — to spot a recycled breach sold as new, match duplicate threads across forums, cluster threat-actor aliases into one persona, and map a campaign over time. Each correlation saves a defender time, money and panic. Defensive, vendor-neutral, with a free synthetic training lab. Credits Iván Castañeda (@ivancastl) / darkforums-ip-intel as real-world prior art.

Eight eMudhra intermediate CA keys were reported compromised (Oct 2026). Browsers revoked within days — curl, containers and language trust stores didn't. The real gap a CA compromise exposes, and how to check yours.

Anthropic’s Claude API and SDK can now emit OpenTelemetry spans for Messages API requests, tool uses and Managed Agents sessions (beta) — to any OTLP backend. For security and audit leaders that is the missing evidence layer: you cannot audit agent behaviour you never recorded. A third-party, docs-grounded read of what it traces (real span + attribute names), why it matters for AI audit and detection, and the one new data-exposure surface it opens — full-content spans that put prompts, tool results and PII into your observability stack.

OVHcloud disclosed (19 October 2026) that its Command Line Interface repository was briefly compromised in a supply-chain attack: per CEO Octave Klaba, an attacker used a developer’s GitHub account to add a binary that steals secrets from disk. Malicious versions v0.11.0–v0.15.0 were blocked ~94 minutes after a customer flagged them; v0.10.0 and v0.16.0 are clean. A third-party, figures-attributed read of what happened and the release-pipeline and developer-account controls every security leader should check — with the facts attributed to The Stack and OVHcloud’s public statements.

KrebsOnSecurity reports (19 October 2026) that U.S. federal investigators suspect an executive at a U.S. consulting firm helped run the ShinyHunters extortion site "shinywire[.]com," and that investigators identified executives at two ransomware-negotiation firms with "unusually close relations" to ShinyHunters. Nothing here is proven against any named individual, and the suspected site-operator is unnamed and uncharged. For CISOs and IR leaders the lesson is structural: the negotiators, IR vendors and "recovery" brokers you bring in mid-incident are themselves a trust boundary. An independent, attributed read on what is alleged versus proven, and how to run conflict-checked, sanctions-screened due diligence on anyone you let into a live ransomware response.

Check Point reports a fast-growing phishing campaign that abuses legitimate calendar features — malicious .ics attachments and fake Microsoft Teams invites — to steal Microsoft 365 credentials, with more than 302,386 unique phishing emails hitting over 40,000 organizations and volume climbing from a few thousand a month in April to more than 740,000 in the first two weeks of October. This is not a flaw in Outlook, Teams or the .ics format; it is the abuse of trusted calendar behaviour. A third-party, figures-attributed read of what the campaign is, why calendar invites bypass normal phishing instincts, and the defensive playbook for security leaders. Every number is attributed to Check Point.

A security-leadership field guide to the AI-agent benchmarks vendors quote: WorkBench, JobBench, BigLaw Bench (+ BigLaw Bench: Research), SWE-bench / SWE-bench Verified, GAIA, τ-bench / τ²-bench, WebArena, MLE-bench, BrowseComp and AssistantBench. What each one measures, how grading differs (programmatic vs rubric vs human), why a high score on one says almost nothing about another, and the questions to ask when a benchmark number lands on your desk. Tied to Google’s Harness² work, which is a reminder that a score is the model plus its harness — not the model alone.

A security-leadership comparison of the main ways to make an AI agent better without retraining the model: Google’s Harness² (recursive harnessing of the whole scaffold), GEPA (reflective prompt evolution), DSPy (a pipeline compiler with optimisers), ADAS (a meta-agent that writes new agents in code), Reflexion (runtime verbal self-reflection) and plain ReAct as the baseline. What each optimises — prompts, few-shot demos, tools, topology, control flow or memory — the data and cost each needs, its maturity, and the governance implications (self-modifying agents, code generation, egress, auditability). Includes an honest note on "PRISM," a name that maps to several unrelated methods.

Google’s Harness² lets an AI agent rewrite its own harness — prompts, guardrails, memory, tools — per task, weights fixed. A calm read of what it is and the governance controls it demands.

ISO/IEC 27090:2026 is the first international standard for AI security threats. What it covers, why there is no "ISO 27090 certificate," and where to get the official PDF.

OpenCode (Anomaly Innovations) notified a subset of users on 17 Oct 2026 that an intruder reached a legacy system and may have exposed stored OpenAI, Anthropic and Google API keys; developers report thousands in fraudulent inference spend. A calm, third-party read of what is confirmed, what is not, and exactly what to do if you use any bring-your-own-key AI tool. PlayCISO is not affiliated with OpenCode; every figure here is attributed.

Booz Allen’s OT Cybersecurity Lab ran two unnamed frontier AI models against a model manufacturing plant — real PLCs, HMIs, a SCADA platform, a VFD/motor and a robotic arm. Across 8 controlled scenarios every objective was met; one run reached the control network in ~16 minutes; a model took admin control of a robotic arm and moved it. But a human approved every exploit and every physical action, some controls held, and the Cloud Security Alliance calls it a vendor-reported lab finding, not evidence of autonomous AI attacks in the field. A calm read of what it means for OT defenders — and what to do first.

A neutral, decision-framework comparison of four cyber-insurance carriers — Coalition and At-Bay (technology-led "active" insurtechs) versus AIG CyberEdge and Chubb (large traditional carriers). How they differ on underwriting and active monitoring, incident-response panels, ransomware and cyber-extortion coverage, pricing drivers, and healthcare-specific considerations. No invented prices or sublimits — a framework to take to a broker, plus the exact clauses and questions to verify on a quote. Not insurance advice; confirm current terms with a licensed broker.

A 2026 Semperis ransomware report, covered by The Register, finds 37% of ransomware attacks now involve psychological or physical threats against staff — doxxing, swatting, threats to families, even packages sent to investigators' homes. Here is what the numbers say, the caveats that come with a self-reported vendor survey, and what the shift from data-encryption to human coercion changes for incident-response planning and a security leader's duty of care.

Researchers at swarmcha.se report that between 28 Sept and 8 Oct 2026, a fleet of internet-connected AI agents running on Tencent Cloud repeatedly tried to extract Amap (Alibaba mapping) data — bypassing anti-bot defences by abusing a public URL-scanning service as a remote browser, and in several cases using exposed API credentials found in public source. Attribution to any specific Tencent team is explicitly uncertain. Here is what was reported, the heavy caveats the authors themselves raise, and what it means for anyone running or defending against internet-connected AI agents.

A threat researcher reports finding a live, custom-built attack-orchestration platform — "Cairn" — that lets an LLM run intrusions end-to-end with no human in the loop per session: one Docker/Kali container per victim, a fact-graph that remembers what was tried, 229,828 AI-dispatched attack events across 13 days, and a ShellShock-over-SMTP vector that actually landed. Here is what was reported, what is genuinely new (a model reading its own prior work), the IOCs and CVEs involved, and the defences that matter — treated as a single-source disclosure, not yet corroborated by major vendors.

New research from Anthropic’s Frontier Red Team argues that small open-weight AI models are months, not years, from being able to self-replicate across ordinary consumer hardware — no data centre required. The team ran 9B-parameter inference on cheap Android TV boxes and other IoT devices. Here is what the research actually found (and what it did not), why it reframes AI risk as an epidemic-response problem, and the concrete defences — network-anomaly detection, honeypots, ISP quarantine, killing default passwords — that security leaders can act on now.

A cyberattack rarely bankrupts a company via the ransom itself — it is the cascade: operations freeze, customers flee, remediation and notification costs pile up, lawsuits and regulators arrive, and lenders pull out, all at once. Here is a cited analysis of real companies that filed Chapter 11 or collapsed after a breach — AMCA, National Public Data, KNP Logistics, Stoli, ENGlobal — the pattern that killed them, who is most at risk, and what actually prevents it.

Within days, two European grocery giants were hit by very different cyber incidents that ended the same way — empty fresh-food shelves. Albert Heijn (Ahold Delhaize) took its websites, apps and logistics systems offline after a cybersecurity incident, halting online deliveries and stalling distribution centres; in France, attackers hijacked Danfoss refrigeration controllers and cooked the fridges. One is an IT/logistics failure, the other OT sabotage. Here is what is confirmed, how they differ, and why food retail is now critical infrastructure.

On 11–12 October 2026, attackers hijacked refrigeration controllers in hundreds of French supermarkets and forced defrost cycles that heated units toward 50°C, spoiling food across Intermarché, Carrefour, Monoprix, Leclerc, Coopérative U and Super U. CERT-FR confirmed compromises of Danfoss AK-SM and AK-SC controllers and told operators to disconnect them. Here is what is confirmed, what is still unverified, why refrigeration is OT, and the free tools to check your own exposure.

CVE-2026-85097 is a maximum-severity (CVSS 10.0) unauthenticated arbitrary-file-upload flaw in Bricksforge, a popular add-on for the Bricks WordPress builder, that leads to remote code execution. It is being exploited in the wild from October 7, 2026. Here is exactly how the attack works, why patching alone is not enough, and the defender’s checklist — patch to the latest release, hunt the indicators of compromise, and assume exposed sites may already be breached.

Abliteration removes an open-weights LLM’s ability to refuse by deleting its "refusal direction" — not a jailbreak or a retrain. How to govern the risk.

REA (Reverse Engineer Anything) is an open-source GitHub MCP server for AI-driven reverse engineering of apps and binaries. What the repo does, how to use it, and the defender's take.

What Google AI Edge Foresight is: Google's on-device, local-first AI meeting notetaker for Apple Silicon Macs — plus the platform/Windows question and the security checks a CISO runs first.

A dark-web monitoring account reports that the ransomware group "The Gentlemen" has named Deloitte on its extortion leak site, with the entry marked "Data info — soon." As of now this is an unverified claim: a listing appeared, no data or proof has been published, and Deloitte has not confirmed any incident. Here is what "being named" on a leak site actually means, why even a Big Four audit-and-cyber firm appearing matters, how to read these claims without panicking, and the exact playbook if you are a Deloitte client or sit in their supply chain.

A16z, using Epoch.ai data across 21 major software vendors (Apple, AWS, Microsoft, Google, Cisco, Oracle and more), shows critical-severity CVEs that never cleared ~100 a month for four years suddenly jumping past 600 a month since spring 2026; high-severity reports spiked toward ~2,200 a month. In parallel, Kevin Mandia (Mandiant’s founder) says his new venture Armadin has found 90+ zero-days in production at customer sites this year using models post-trained with real red-teamers — black-box, from the internet, no source code. What’s confirmed vs. reported, why the curve bent, and what a security leader should do now.

An independent, attributed roundup of the major cybersecurity incidents of September 2026 — from the actively-exploited Citrix NetScaler and Arista VeloCloud zero-days to the Pentagon Defense Manpower Data Center breach, the Gyazo compromise, the Keio ransomware attack and the disputed ShinyHunters FBI-jobs claim. Each item carries a verification status (confirmed, claimed, reported or rumoured) and a clearly-labelled subjective editorial severity, with a defensive takeaway.

Comparing cyber insurers is hard because pricing is quote-specific and coverage is full of fine print. The dimensions that actually differentiate carriers, and how to run a fair comparison.

Coalition vs At-Bay cyber insurance: both tech-driven "active" insurers. A neutral framework to compare them for your organisation and get real quotes.

Coalition vs Corvus cyber insurance: two tech-driven insurers that pair coverage with security data. A neutral framework to compare and get real quotes.

Coalition vs Resilience cyber insurance: both tech-driven "active" insurers. A neutral framework to compare the dimensions that matter and get real numbers.

Is Coalition cheaper than AIG (CyberEdge) for cyber insurance? It depends on your risk profile. Compare the two on cost, coverage and claims — and how to get real quotes.

Chubb vs Coalition cyber insurance: a neutral framework to compare coverage, incident response and pricing — and get real quotes from both.

A practical vendor security assessment checklist: the areas to review before onboarding a third party, the evidence to ask for, and how to rank vendors by risk instead of treating them all the same.

Security questionnaires are the backbone of third-party risk management — when they are focused. What to include, how to avoid questionnaire fatigue, and how to turn answers into decisions.

A PCI DSS gap analysis compares your current controls against the standard's requirements before a formal assessment. How to scope it, run it, and turn the findings into a remediation plan.

Annual Loss Expectancy (ALE) puts a dollar figure on cyber risk. The ALE = SLE × ARO formula, a worked example, and a calculator to run your own numbers.

SLE is the cost of one incident; ALE is the expected cost per year. How SLE, ARO and ALE connect, the formulas, and when to use each in risk analysis.

A vendor-neutral ransomware readiness assessment: the controls that actually decide whether an incident is a bad day or a business-ending one, grouped so you can score yourself and find the gaps.

Data breach cost is more than the ransom or fine. A breakdown of the direct and indirect costs, what moves the number, and how to estimate yours.

A practical walkthrough of how to run a NIST CSF 2.0 assessment: the six Functions, the four Implementation Tiers, self-assessment vs. formal review, common pitfalls, and how to turn the result into a roadmap your board will approve.

Free, browser-based CISO training games — no signup, no tabletop scheduling. Practice phishing judgment, MFA-bombing triage, threat-actor ID, and boardroom pitches in minutes.

An independent, defensive roundup of August 2026’s biggest cyber incidents: the BeyondTrust Remote Support zero-day (CVE-2026-102764) exploited by a China-nexus actor from late August, the McKesson and Manchester Airports Group breaches, a record 997 ransomware claims, social-engineering hits on Apollo, RingCentral and Carhartt, OT attacks on energy and water, and the AI angle. Each item carries a status label (Confirmed / Claimed / Reported) and a subjective 1–10 severity read.

A practical roadmap for security leaders preparing to step into the CISO seat — from closing executive skill gaps to practicing board-level decisions under pressure, with the tools and frameworks that actually build readiness.

A practical comparison of the platforms security leaders are using in 2026 to develop executive skills, practise board communication, and prepare for the CISO role — from simulation-based training to certifications and peer networks.

A comparison of the interactive incident response simulators available in 2026 for security leaders preparing for CISO and executive roles — from AI-driven War Room scenarios to tabletop exercise platforms and cyber range tools.

CISO candidates need to practise breach response at the executive level — disclosure decisions, regulatory notifications, board communication, and crisis management — before they are responsible for doing it for real. Here is how to build that muscle.

An independent, defensive roundup of the major cyber incidents reported in July 2026 — the Fairlife/Coca-Cola ransomware shutdown, Clop's PTC Windchill data-theft campaign, the 12-million-record KDDI breach, Accenture's source-code incident, a coordinated OT attack on 30+ Minnesota water utilities, and the actively-exploited ColdFusion and Langflow flaws. Each item carries a status label (Confirmed / Claimed / Reported) and a subjective 1–10 severity read, with a defensive takeaway.

An independent, attributed roundup of the major cybersecurity incidents of June 2026 — from the Oracle PeopleSoft zero-day (CVE-2026-35273) that ShinyHunters used against 100-plus universities and Microsoft’s record ~206-CVE Patch Tuesday to breaches at iRhythm, Novo Nordisk, Tata Electronics and Kodak and a wave of npm supply-chain compromises. Each item carries a verification status (confirmed, claimed or reported) and a clearly-labelled subjective editorial severity, with a defensive takeaway.

An independent, attributed roundup of the major cybersecurity incidents of May 2026 — from the ShinyHunters Canvas/Instructure extortion and the TanStack supply-chain worm that reached OpenAI to the DigiCert code-signing abuse, a suspected Iranian attack on US gas-station tank gauges, and breaches at Carnival, Charter/Spectrum, West Pharmaceutical and more. Each item carries a verification status (confirmed, claimed or reported) and a clearly-labelled subjective editorial severity, with a defensive takeaway.

An independent, attributed roundup of the major cybersecurity incidents of April 2026 — from the ShinyHunters Salesforce extortion wave hitting Charter/Spectrum, ADT, Pitney Bowes, Vimeo and Medtronic, to the ChipSoft ransomware attack on Dutch hospitals, the France Titres (ANTS) identity-portal breach, a Microsoft SharePoint zero-day and the ~$290M Kelp DAO crypto theft. Each item carries a verification status (confirmed, claimed or reported) and a clearly-labelled subjective editorial severity, with a defensive takeaway.

An independent, attributed roundup of the major cybersecurity incidents of March 2026 — from the TeamPCP supply-chain campaign that poisoned LiteLLM and security-scanner packages on PyPI, to the Handala wiper that erased 200,000+ Stryker devices, the Cl0p Oracle E-Business Suite extortion wave, and breaches at Cegedim Sante, Navia, Aura, Crunchyroll and Telus. Each item carries a verification status (confirmed, claimed or rumoured) and a clearly-labelled subjective editorial severity, with a defensive takeaway.

An independent, attributed roundup of the major cybersecurity incidents of February 2026 — from Microsoft’s six actively-exploited zero-days and the BeyondTrust and Ivanti EPMM flaws to breaches at the University of Mississippi Medical Center, France’s national bank account registry (FICOBA), CarGurus, Figure, Advantest and Singapore’s four telcos. Each item carries a verification status (confirmed, reported or claimed) and a clearly-labelled subjective editorial severity, with a defensive takeaway.