Cybersecurity Executive Brief: March 2026

Section 1 — Executive Brief #

Cybersecurity Insights for Decision-Makers | 2-Minute Executive Summary | Full Analysis: ~10 pages

Scope: Incidents disclosed or materially updated during March 2026.

Three Assumptions That March Exposed #

The most dangerous software in your environment isn’t malware. It’s the security tool you never thought to verify.

If your organization depends on open-source tooling — and it does — March proved that trust without verification is a liability your board can no longer afford to ignore. Here are three ASSUMPTIONS that March 2026 exposed.

In November 2018, a developer named Dominic Tarr received a simple message on GitHub: “Can I help maintain this?”

The “this” was event-stream — a JavaScript library with 1.5 million weekly downloads. Tarr had lost interest in maintaining it. The offer seemed reasonable, even generous. He handed over the keys.

The new maintainer waited a few weeks, then slipped a malicious dependency into the package. It contained an encrypted payload designed to steal cryptocurrency wallet keys. The backdoor sat in the npm registry — the public repository where JavaScript libraries are shared — for two months before anyone noticed. It was a single developer who spotted an unexpected warning during a routine build.

No exploit. No vulnerability. Just a polite request and an implicit assumption: that the person maintaining a tool millions of developers depend on had been properly vetted. They hadn’t. Nobody’s job description included checking.

That was eight years ago. The lesson should have reshaped how the industry thinks about trust in open-source dependencies. For most organizations, it didn’t. In March 2026, that same implicit trust — scaled to security tooling, automated build pipelines, and AI infrastructure — produced what may be the most consequential open-source supply chain attack the industry has seen.

Assumption #1: The tools that scan for vulnerabilities are themselves secure #

On March 19, a threat group called TeamPCP exploited an incomplete credential rotation to poison Trivy — the most widely used open-source cloud vulnerability scanner.¹ Not a random library. The scanner your security team likely runs to find vulnerabilities in your container images. TeamPCP pushed malicious code across 76 of 77 version tags. The payload harvested cloud credentials, SSH keys, and cloud infrastructure secrets from every automated build pipeline that ran it.

Within eight days, the compromise cascaded to Checkmarx KICS (another security scanning tool), LiteLLM (an AI gateway with 95 million monthly downloads), Telnyx (a communications platform), and 47 additional npm packages.²

An AWS secret stolen through the Trivy compromise was used to access the European Commission’s cloud environment. CERT-EU confirmed that about 91.7 GB was exfiltrated from a compromised AWS account supporting websites for 42 Commission clients and at least 29 other Union entities. ShinyHunters published the dataset on March 28.³

Google’s Threat Intelligence Group warned that these compromises could leave very large numbers of secrets circulating, enabling follow-on SaaS compromise, ransomware, extortion, and additional supply chain attacks.⁴

The tool the European Commission trusted to find vulnerabilities was the vulnerability. Security tools sit at the most privileged point in your pipeline — they have access to source code, build secrets, and deployment credentials by design. When one is compromised, it becomes the threat, with access that monitoring tools won’t flag because the tool is supposed to be there.

Assumption #2: A compromised dependency will be caught before it reaches production #

On March 31, a North Korea-nexus actor compromised the Axios npm package — the most popular HTTP client in JavaScript, with over 70 million weekly downloads.⁵ Backdoored versions containing a cross-platform remote access tool were live for approximately three hours before detection. Google’s Threat Intelligence Group confirmed this was independent from TeamPCP — meaning two separate actors targeted the same open-source ecosystem in the same month.⁶

Three hours. For organizations with automated build pipelines that pull the latest version without pinning, three hours is enough.

The TELUS Digital breach reinforced the same lesson from a different angle. Reuters and TELUS confirmed that ShinyHunters breached the Canadian BPO provider; ShinyHunters claimed they gained access through Google Cloud credentials found in data stolen during the 2025 Salesloft Drift breach.⁷ If accurate, the initial compromise wasn’t technical — it was a failure to rotate credentials after a prior incident.

The assumption that “someone would catch it” is no longer adequate. Dependency integrity must be verified at consumption, not assumed at publication.

Assumption #3: Supply chain risk is a procurement problem, not an operational one #

Most enterprise risk frameworks treat supply chain as a vendor management exercise — questionnaires, SOC 2 reports, contractual obligations. March proved that the operational supply chain — the code your developers pull into every build, the AI gateways routing your LLM traffic — carries strategic risk that no vendor questionnaire will surface.

TeamPCP didn’t breach a vendor. They poisoned tools that vendors and enterprises alike trust without question. The European Commission wasn’t compromised through a negotiated contract — it was compromised because a vulnerability scanner in an automated pipeline was quietly replaced with a weaponized version.

The same blind spot extends to the platforms enterprises use to manage their own infrastructure. Geopolitical risk became operational for Stryker Corporation. Stryker’s SEC 8-K confirmed a March 11 cyber incident that disrupted its Microsoft environment. A March 16 customer certification letter stated that the threat actor used Microsoft Intune to wipe internal virtual infrastructure. Iran-linked group Handala claimed responsibility, asserting they destroyed over 200,000 systems.⁸ Manufacturing was disrupted globally. Surgical procedures were delayed. CISA and the FBI engaged, alongside broader U.S. government engagement.

And the Tycoon 2FA takedown — a six-nation operation that seized 330 phishing domains — revealed how industrialized credential theft has become: 62% of all phishing blocked by Microsoft, 30 million emails per month, sold as a subscription for $350.⁹ CrowdStrike observed phishing volumes return to pre-disruption levels within days.

The line to remember: March didn’t reveal a new category of threat. It revealed how deep the old assumption runs — that the tools and partners you depend on have already been validated by someone. In most cases, they haven’t.

Critical Actions Required #

🔴 IMMEDIATE — Ask your CISO whether your CI/CD pipelines — the automated systems that build and deploy your software — verify dependency integrity or pull the latest version automatically. The TeamPCP campaign succeeded because pipelines trusted whatever version was tagged as current. Require a review of version pinning, lockfiles, and integrity verification for all production builds.

🔴 THIS WEEK — Confirm that all secrets exposed to CI/CD environments have been rotated. GTIG warned that very large numbers of stolen secrets may be circulating. Any organization that ran Trivy, KICS, LiteLLM, or the 47 affected npm packages should treat all pipeline secrets as potentially compromised.

🟡 THIS MONTH — Review your software supply chain governance. Ask your CISO: “Do we have an inventory of every open-source tool that runs with privileged access in our build pipeline? Who validates their integrity?” March demonstrated that SOC 2 reports don’t cover the dependency graph.

🟡 THIS MONTH — Confirm that device management and disaster recovery infrastructure has been assessed for compromise. The Stryker incident demonstrated that management tools are being targeted specifically because they have the authority to destroy at scale.

🟢 MONITOR — The U.S. government released a new cyber strategy on March 6, prioritizing offensive operations.¹⁰ The latest Regulatory Agenda targets May 2026 for the CIRCIA final rule (the pending federal incident-reporting mandate). The CISA 2015 Information Sharing Act has been extended through September 30, 2026, but faces uncertain reauthorization. If your threat-sharing posture depends on these protections, track the renewal timeline.

🟢 MONITOR — Breachsense tracked 808 organizations listed on ransomware leak sites in March, a 19% month-over-month increase. Q1 2026 totaled 2,165, putting the year on pace for 8,660 — up 18.5% over 2025.¹¹ Identity-first compromise has become the leading entry vector.

Key Metrics #

What to Say When the Board Asks… #

“Are we exposed to these supply chain attacks?” “Every organization that uses open-source software is — which means every organization. We’re [reviewing / have reviewed] our CI/CD pipeline to confirm dependency integrity, and rotating all secrets that may have been exposed. We’re also inventorying which security tools run with privileged access in our build pipeline.”

“What happened at Stryker?” “Stryker disclosed a cyber incident on March 11. A follow-up customer letter confirmed Intune was used to wipe internal infrastructure. An Iran-linked group claimed responsibility. The question for us is whether our administrative access controls would prevent a single compromised account from executing mass destructive actions through our device management platform.”

“Should we be worried about the ransomware numbers?” “808 organizations were listed on leak sites in March. The trend is directional: volume is up, and attackers increasingly use stolen credentials rather than exploits. Our priority is identity monitoring — detecting bulk exports, impossible-travel logins, and MFA changes in real time.”

Forward Section 2 to your CISO with one question: “What are we doing about each of these?”

Section 2 — Full Analysis #

Why March Matters #

March 2026 is the month the software supply chain became a battleground — and the weapons were the defenders’ own tools.

Two independent campaigns ran against the open-source ecosystem in the same month. TeamPCP poisoned Trivy, the most widely adopted vulnerability scanner, and cascaded to four additional tools in eight days. The European Commission was breached through its own security pipeline. Separately, a North Korea-nexus actor (Microsoft: Sapphire Sleet; GTIG: UNC1069) compromised Axios, deploying a cross-platform backdoor to a package with over 70 million weekly downloads. GTIG confirmed the campaigns were independent (see notes 5–6).

Three dimensions matter for security leaders:

The attack surface moved to security infrastructure itself. Trivy sits at the center of the trust model, not the edge. Compromising it gives access to the secrets that secure everything else.

The blast radius extends beyond code. TeamPCP harvested CI/CD secrets, and Wiz observed those secrets being quickly validated and used to explore victim environments. Supply chain compromise is no longer about injecting malicious code — it’s about weaponizing trust relationships in automated pipelines.

The speed of cascade defeats traditional detection. Nine days from initial Trivy compromise to the Commission’s data being published. The Commission’s CSOC received its first alerts on March 24 — days after the compromise began. Automated pipelines accelerate the propagation of compromise as efficiently as they accelerate development.

Critical Incidents: Top 5 #

1. TeamPCP Supply Chain Campaign (Trivy → KICS → LiteLLM → Telnyx → EU Commission) #

What happened: By scope and downstream impact, arguably the most significant open-source supply chain attack since SolarWinds. TeamPCP exploited incomplete credential rotation from a February 2026 breach of Aqua Security to force-push malicious code to 76 of 77 Trivy version tags. The payload harvested SSH keys, cloud credentials, and Kubernetes secrets from CI/CD runner memory. The cascade: KICS (35 tags, March 21), Aqua’s full GitHub org (44 repos defaced, March 22), LiteLLM on PyPI (March 23), 47+ npm packages via CanisterWorm, and Telnyx via WAV steganography (March 24). CVE-2026-33634 (CVSS 9.4).¹²

The European Commission’s AWS environment was accessed using a stolen secret. CERT-EU confirmed about 91.7 GB exfiltrated from a compromised AWS account supporting websites for 42 Commission clients and at least 29 other Union entities. GTIG warned the compromises could leave very large numbers of secrets circulating, enabling follow-on SaaS compromise, ransomware, extortion, and additional supply chain attacks. Unit 42 reported that TeamPCP announced a partnership with the Vect ransomware group on BreachForums to monetize the stolen credentials.

The LiteLLM compromise is particularly notable: the malicious .pth file auto-executed on any Python process on the host, not just when LiteLLM was imported — broadening the blast radius to any application on the same machine.

What to do:

• CEO/CFO: Treat software supply chain risk as a board-level topic — comparable to third-party vendor risk. Ask: “Do we have an inventory of every tool with privileged access in our build pipeline, and who validates their integrity?”
• CISO: Rotate all secrets exposed to CI/CD environments. Mandate version pinning and lockfiles. Implement SBOM generation and integrity verification for security tooling. Conduct host-level forensics for any system that ran compromised LiteLLM versions.
• Legal/Compliance: Assess notification obligations if CI/CD secrets included customer data access credentials. GTIG’s assessment on circulating secrets could raise questions about proactive breach assessment requirements under DORA, NIS2, or SEC disclosure rules.

Full sourcing in notes 1–4 and 12.

2. Stryker Corporation — Wiper Attack (Handala/Iran) #

What happened: Stryker Corporation disclosed a cyber incident via SEC 8-K on March 11. The filing confirmed disruption to its Microsoft environment. A March 16 customer certification letter stated that the threat actor used Microsoft Intune to wipe internal virtual infrastructure. Iran-linked group Handala claimed responsibility, asserting they obtained Global Administrator access and destroyed over 200,000 systems, with 95% of devices erased in some departments. Handala claimed 50 TB exfiltrated. Neither the system count nor exfiltration volume has been independently verified. Specops/Outpost24 identified 278 compromised stryker.com credentials between October 2025 and March 2026 (see note 8).

The broader Iran-cyber conflict escalated in parallel: U.S. Cyber Command disrupted Iranian communications during strikes, compromised Tehran traffic cameras, and hacked the BadeSaba prayer app. Israeli forces bombed the IRGC cyber HQ. CCCS issued advisories for pro-Iran PLC targeting campaigns (Operation Olalampo).¹³

Why it matters: What Stryker’s own disclosures confirm is consequential enough: Intune — a tool designed to manage endpoints — was used to destroy them. Manufacturing, order processing, and surgical supply chains were disrupted globally. CISA, FBI, and H-ISAC all engaged, alongside broader U.S. government engagement. This appears to be one of the first significant Iranian cyber retaliations in the current conflict — and it targeted a healthcare supply chain company, not a military installation.

What to do:

• CEO/CFO: Ask whether your device management platform could be used destructively by an attacker with admin access — and what controls prevent it.
• CISO: Implement multi-person authorization for mass-deployment actions. Review CCCS advisories CF26-004 and CF26-005 for Operation Olalampo indicators. Assess geopolitical targeting risk by sector.
• Legal/Compliance: Confirm whether your cyber insurance covers state-attributed destructive attacks — Lloyd’s “act of war” exclusions may apply.

Full sourcing in notes 8 and 13.

3. TELUS Digital / ShinyHunters ($65M Demand) #

What happened: Reuters and TELUS confirmed that ShinyHunters breached TELUS Digital, the BPO arm of Canadian telecom TELUS. ShinyHunters claimed nearly 1 petabyte stolen, alleging initial access via GCP credentials from the 2025 Salesloft Drift breach, lateral movement via TruffleHog across BigQuery instances, and 28 downstream clients affected. The access chain and downstream scope remain attacker claims. ShinyHunters demanded $65 million; TELUS refused. Data was not published as of mid-March (see note 7).

Why it matters: The confirmed breach alone is significant — TELUS Digital serves banking, healthcare, and e-commerce clients. If the claimed credential chain is accurate, it illustrates long-tail supply chain risk: credentials from a prior vendor breach, never rotated, enabling compromise of a BPO provider months later.

What to do:

• CEO/CFO: Ask which BPO providers access your data environments and when their credentials were last audited.
• CISO: Validate that credential rotation policies are enforced after every vendor breach disclosure — not just documented.
• Legal/Compliance: Review BPO contracts for breach notification timelines that cover upstream vendor breaches.

Full sourcing in note 7.

4. Tycoon 2FA — Scale Revealed, Takedown, and Resurgence #

What happened: Europol coordinated with six countries and eleven private sector partners to take down Tycoon 2FA on March 4, revealing its scale: ~2,000 subscribers, 24,000+ domains since launch, 62% of all phishing blocked by Microsoft, 30 million emails per month targeting 500,000+ organizations. Pricing: $120/10 days, $350/month. Microsoft filed a civil complaint against the developer, seeking $10 million (see note 9).

CrowdStrike observed volume drop to 25% on March 4–5. Activity returned to pre-disruption levels within a brief period.

Why it matters: A single platform generated nearly two-thirds of all phishing Microsoft blocked. MFA bypass is a commodity subscription. And 330 domain seizures across six countries produced a temporary dip, not a structural disruption. If your MFA relies on push notifications or SMS, it is systematically bypassable at this scale.

What to do:

• CEO/CFO: Fund phishing-resistant MFA (FIDO2/WebAuthn) for all user populations, starting with administrators.
• CISO: Deploy phishing-resistant authentication for privileged accounts immediately. Monitor for session token theft and AitM indicators.
• Legal/Compliance: For NYDFS-regulated entities, Section 500.12 (effective November 2025) favors higher-assurance MFA methods.

Full sourcing in note 9.

5. Ransomware Volume Acceleration — Q1 2026 #

What happened: Breachsense tracked 808 organizations listed on leak sites in March (65 groups, 75 countries) — a 19% month-over-month increase. These are leak-site claims, not all independently confirmed, but the trend is consistent across sources. Q1 total: 2,165 (annualized 8,660, +18.5% vs 2025). Qilin posted 131 — the highest single month for any group. US accounted for 50% of claims. Manufacturing most targeted (76), followed by construction (53) and finance (48) (see note 11).

Chainalysis reported more than $820 million in on-chain ransomware payments in 2025, noting the total is likely to approach or exceed $900 million as more events are attributed. Per-incident revenue is declining, but volume offsets it. Identity-first compromise has become the leading entry vector.

Why it matters: Lower payments aren’t slowing ransomware — they’re driving it to scale. Identity-first entry is cheaper and faster than exploit development. CrowdStrike reported 29-minute average breakout time (down from 48 in 2025), with exfiltration observed within four minutes in one case.¹⁴

What to do:

• CEO/CFO: Can your organization detect and contain a credential-based intrusion within 30 minutes? If not, pre-staging response is a budget priority.
• CISO: Shift detection to credential abuse indicators: impossible travel, bulk exports, MFA re-enrollment. Validate backup isolation from production credential stores.
• Legal/Compliance: Confirm insurance covers data-theft-only extortion and that response retainers are in place.

Full sourcing in notes 11 and 14.

Other Significant Incidents #

• European Commission cloud breach (March 19–28) — Trivy supply chain; CERT-EU confirmed about 91.7 GB exfiltrated from a compromised AWS account (42 Commission clients, 29+ Union entities); second breach of 2026. (See Incident #1 for full sourcing.)
• Sears AI chatbot — 3.7M chat logs and 1.4M audio files exposed via misconfigured cloud database. AI tool data sprawl as emerging risk.
• Lloyds Banking Group — Customer data sent to wrong recipients (448K customers). Privacy incident, not intrusion.
• Ericsson — Third-party breach from April 2025 disclosed eleven months later. 15K+ affected.
• Interlock ransomware exploited Cisco FMC (CVE-2026-20131) as a pre-disclosure zero-day — confirming the pattern of targeting security management infrastructure.

Key Threat Actors #

Attack Methods: What Your Teams Should Brief Up #

• Security tooling supply chain poisoning. TeamPCP poisoned tools inside the trust boundary. Trivy, KICS, and LiteLLM had legitimate, privileged access to CI/CD secrets. Compromising the verifier, not the thing being verified.
• Dependency maintainer account takeover. Both Trivy and Axios compromises began with control of a legitimate maintainer account. No exploit — just credential access.
• Management plane weaponization. Intune (Stryker), VMware Aria Operations, BeyondTrust Remote Support, Cisco FMC — attackers systematically targeting management, monitoring, and protection tools.
• Commodity MFA bypass. Tycoon 2FA: 62% of Microsoft-blocked phishing. Subscription-priced. Push-based MFA is functionally compromised at the ecosystem level.
• Credential chain exploitation. TELUS Digital reportedly breached via year-old unrotated credentials from a prior vendor breach. Third-party credential chains are a primary attack vector.
• Geopolitically motivated destruction. Stryker wiper, Cyber Command operations against Iran, Operation Olalampo PLC targeting. Enterprise continuity now directly exposed to geopolitical conflict.

Sector Impact Analysis #

Healthcare & Life Sciences: The Stryker wiper attack — disclosed via SEC filing, with a customer letter confirming Intune was used destructively and Handala claiming 200K+ systems destroyed — marked one of the first major Iranian cyber retaliations targeting a healthcare company. Healthcare ransomware dropped 49% from February (93 to 47), suggesting February was campaign-driven. But the shift to operational destruction changes the sector’s risk calculus.¹⁵

Technology & Software Supply Chain: The month’s defining events. Trivy, KICS, LiteLLM, Telnyx, 47+ npm packages, Axios — all compromised. European Commission breached through its own tooling. TELUS Digital demonstrated BPO-as-attack-surface. SentinelOne documented AI EDR autonomously blocking an AI-agent-initiated supply chain attack.¹⁶

Government & Public Sector: European Commission breached for the second time in 2026. U.S. released cyber strategy and EO 14390. State Department launched Bureau of Emerging Threats. Gen. Rudd confirmed to lead CyberCom/NSA. But CISA lacks a confirmed director, the CIRCIA final rule is targeted for May 2026, and the 2015 Information Sharing Act faces September expiration.¹⁷

Financial Services: Finance ransomware rose 30% (37 to 48). DORA submissions continued with some rejected. Intesa Sanpaolo fined for insider access. Insurance market: Munich Re and Gallagher noted sharper “act of war” exclusions and MFA-dependent underwriting.¹⁸

Manufacturing & Industrial: Most targeted ransomware sector for the sixth consecutive month (76 victims). Stryker combined ransomware-scale disruption with geopolitical motivation. Operation Olalampo raised ICS/SCADA risk for Western manufacturers (see notes 8, 11, and 13).

Regulatory Timeline #

Strategic Threat Signals #

The open-source trust model needs reinforcement, not abandonment. March’s attacks indict the way enterprises consume open source, not open source itself. Implicit trust, auto-fetching latest versions, and absent integrity checks for security tooling are governance failures. The fix: SBOMs, pinned versions, signed packages, privileged-tool auditing.¹⁹

AI infrastructure is now a target. LiteLLM represents one of the first high-profile supply chain compromises to hit AI gateway infrastructure — compromising it means access to API keys, model credentials, and prompt data. SentinelOne documented AI EDR autonomously blocking an AI-agent-initiated supply chain attack — a milestone for defensive automation.²⁰

Geopolitical cyber operations have enterprise consequences. Stryker wiper, Cyber Command operations against Iran, IRGC HQ bombing, Operation Olalampo. Organizations with geopolitical exposure face operational destruction risk, not just data theft. The Stryker incident raised immediate “act of war” insurance questions.²¹

Disruption campaigns buy time, not outcomes. Tycoon 2FA’s takedown produced a brief dip before recovery. Disruption forces migration, not cessation. Permanent reduction requires making identity-first attacks unprofitable at the enterprise level.²²

Market & Industry Intelligence #

• Google/Wiz $32B closed (March 11) — Google’s largest acquisition. Hyperscaler investment in security as competitive differentiator.²³
• 38 M&A deals in March. Investment concentrated in identity governance and AI SOC automation. Notable: Linx Security ($50M Series B), Tenex.AI ($250M Series B).²⁴
• RSAC 2026 (March 23–26) marked the shift from AI copilots to agentic workflows.
• Insurance evolution: Stryker raised “act of war” exclusion questions. Munich Re and Gallagher outlooks noted sharper exclusions and MFA-dependent underwriting (see notes 18 and 21).

Investment Priorities #

Board Discussion Points #

1. “Do we know which open-source tools run with privileged access in our build pipeline — and who verifies their integrity?”

2. “If our vulnerability scanner were compromised, how long would it take us to detect it?”

3. “Which management platforms could be used destructively by an attacker with admin access — and what prevents that?”

4. “Have we rotated all credentials exposed to our CI/CD environment in the past 90 days?”

5. “Does our cyber insurance cover a state-attributed destructive attack?”

6. “Is our MFA push-based or phishing-resistant — and what’s the upgrade timeline?”

Next Month Predictions #

Will happen (high confidence):

• Supply chain attacks continue. TeamPCP remains active; their announced partnership with Vect creates a monetization path for stolen secrets.
• Ransomware volume sustains or increases. Identity-first entry is cheaper than exploitation.
• PhaaS platforms rebuild after Tycoon 2FA takedown.

Might happen (50–70%):

• A major enterprise discloses a breach traced to the TeamPCP campaign. GTIG’s assessment on circulating secrets suggests downstream compromises are likely.
• Additional Iranian cyber operations targeting Western critical infrastructure.
• CIRCIA final rule (May) introduces immediate compliance planning requirements.

Watch for (low probability, high impact):

• A wiper targeting multiple organizations via shared management platform — the Stryker playbook scaled.
• Supply chain compromise of an AI model hosting platform — the LiteLLM compromise showed the pattern; it will repeat.
• Activation of pre-positioned state access in critical infrastructure during geopolitical escalation.

Key Conclusions #

• Your security toolchain is an attack surface. Vulnerability scanners, IaC tools, and AI gateways — trusted by design — were weaponized. Trust without verification is no longer viable.
• The open-source trust model needs verification, not abandonment. March’s attacks are enterprise governance failures, not ecosystem failures.
• Geopolitical conflict now means operational destruction. The Stryker incident made the abstract concrete: enterprise management tooling used to destroy, not steal.
• Identity-first compromise is the dominant pattern. Ransomware, credential theft, supply chain pivots, BPO exploitation — the common thread is identity.
• Disruption campaigns buy time, not outcomes. Tycoon 2FA’s recovery demonstrates that enforcement alone cannot solve the economic model.

Bottom Line #

March 2026 proved that the implicit trust enterprises place in their software supply chain — the tools, dependencies, and automated pipelines that underpin modern operations — is a strategic liability that boards can no longer delegate to engineering teams alone. The organizations that verified before trusting were not on the casualty list.

Sources #