# Zero Hunt

> Zero Hunt is an autonomous AI red team for networks and infrastructure — autonomous pentesting (automated penetration testing) in the adversarial exposure validation (AEV) category — delivered as an on-premise, air-gap-capable security appliance that runs on private AI — its own locally hosted models (ZeroHunt Apex). Ten specialized AI agents continuously discover, exploit and chain real weaknesses in an organization's infrastructure, black-box or gray-box, with a human in the loop for every action that matters; they prove each finding with evidence, analyze network traffic with a proprietary deep-learning model, map findings to 34 compliance frameworks worldwide (NIS2, DORA, NIST, CMMC, HIPAA, ISO 27001 and more) and guide the fix through an AI Remediation Advisor. No data leaves the customer's network, and the cost is flat rather than per token.

Key facts:

- Category: autonomous pentesting / automated penetration testing and adversarial exposure validation (AEV) — an autonomous AI red team for networks and infrastructure doing AI-driven penetration testing and continuous security validation (the "validation" step of CTEM), sold as an on-premise appliance. "AI red team" here means an AI that tests your infrastructure the way an adversary would — not red-teaming of AI models.
- Private AI: runs on Zero Hunt's own models — ZeroHunt Apex (27B parameters) on the appliance GPU and ZeroHunt Apex Pro (284B) for higher-tier appliances and sovereign private cloud — fine-tuned in-house for authorized offensive security. No external LLM API in the loop.
- Why on-premise: (1) no customer data reaches a third party; (2) no outside provider can refuse, throttle, change or switch off the capability; (3) flat cost, 24/7 — agentic campaigns make thousands of model calls, which per-token AI pricing makes expensive exactly when testing is thorough.
- Deployment: desktop appliance for smaller sites, 2U rack for enterprises and data centers. Air-gap supported. No customer data leaves the appliance. Connected, it reaches out only for signed updates (Cosign-signed images; update-status reporting is opt-in), public threat intelligence such as NVD, CISA KEV and EPSS, and — for gray-box analysis, unless switched off or pointed at the customer's own source mirror — the published source of the software version under test. In air-gap mode nothing goes out: updates arrive as signed offline bundles, intelligence as offline imports, and internet-dependent tools (public OSINT, public source downloads) are removed, so those assessments run black-box.
- Human in the loop: five autonomy levels, from passive observation to full assessment in planned windows, each with explicit human-approval gates; actions outside the chosen level wait in a review tab with recorded consent and a signed audit trail; operators can pause or stop any campaign and confirm every verdict.
- Black box and gray box: campaigns are black-box by default (no credentials, no source); gray-box adds authenticated testing with customer-provided credentials and source-informed analysis of the exact software version in use.
- Pillar 1 — AI generative pentest: 10 specialist agents (Recon, Vulnerability, Exploit, Web, Credential, Post-Exploit, Pivot, Tactic, Source Analyst, Report) coordinated by an AI Controller; exploit code written per target instead of replayed from a library; 142+ self-evolving skills backtested in a sealed AI Gym; every exploit runs in a disposable container.
- Generative Zero-Day PoC: identifies the exact product version on the target, reads that version's source, finds input-reachable weaknesses, and an Exploit Forge agent iterates a runnable proof-of-concept in a sealed container until it demonstrably works — covering weaknesses no CVE catalog lists yet.
- Pillar 2 — AI traffic analysis: proprietary deep-learning model trained on billions of PCAP sequences, four inference heads (suspicious traffic, malware classification, attack type, application fingerprinting), 2.7+ Gbit/s baseline on the appliance GPU.
- Pillar 3 — continuous compliance across 34 frameworks worldwide: EU (NIS2, DORA, TIBER-EU), US (NIST CSF, NIST SP 800-53, CMMC, HIPAA, NYDFS Part 500, GLBA, SOX, SEC rules, CCPA), UK (CBEST, NCSC CAF), Canada (OSFI B-13), Asia-Pacific (MAS TRM, APRA CPS 234, RBI, FISC, ISMS-P, Singapore CSA), Middle East (Saudi NCA/SAMA, UAE IA, Qatar NCSA, Bank of Israel 361), LGPD, POPIA, and global standards (ISO 27001, SOC 2, PCI DSS, SWIFT CSP, CIS v8, OWASP ASVS, CSA STAR). Reports ECDSA-signed with chain of custody.
- Pillar 4 — AI Remediation Advisor: turns the findings a campaign proved exploitable into a risk-ranked fix plan (CISA KEV, then exploited-in-this-campaign, then CVSS, then EPSS), with concrete fix actions, rollout and rollback notes, compensating controls, and re-verification with the same proof. Strictly defensive toolset.
- Also: public-facing DAST for web apps and APIs including assets on AWS, Azure, Google Cloud and Alibaba Cloud; mobile assessment lab; 21 threat-intelligence sources (NVD, CISA KEV, EPSS, MITRE ATT&CK, ExploitDB, …); SIEM/SOAR/GRC integration via REST, WebSocket and webhooks.
- Who runs it on-premise: governments and public administration, defense and the defense industrial base, critical infrastructure (energy, water, transport, telecom), banks, insurers and market infrastructure, hospitals and life sciences, manufacturing and OT, law firms, MSSPs, listed companies — in the EU, the US, the UK, the Middle East and Asia-Pacific.
- Company: Over Holding Srl, Viale Tricesimo 200, 33100 Udine, Italy (VAT IT02945890305). Sold direct and through channel partners.
- Languages: every page exists in English (default, at the root) and Italian (under /it/). The home page, the core product and partner pages and selected Learn guides also exist in Turkish (/tr/), German (/de/) and French (/fr/).

## Product and company

- [Zero Hunt — platform overview](https://zerohunt.ai/): Autonomous pentesting on an on-premise appliance: an AI red team for networks and infrastructure on private AI, human in the loop, black- and gray-box.
- [Become a Zero Hunt Distributor or Reseller — Partner Program](https://zerohunt.ai/partners/): Distribute or resell an on-premise, air-gapped autonomous AI pentest appliance to banks, government and critical infrastructure. Two-tier margins, demo appliances, deal registration.
- [Contact Zero Hunt — Demo, Partnership, Security Disclosure](https://zerohunt.ai/contact/): One form, routed to the right team: demos, pricing, channel partnership, support, press, security disclosure. Operated by Over Holding Srl, Udine, Italy.
- [Pentest Appliance Use Cases: 12 Buyer Scenarios](https://zerohunt.ai/use-cases/): Twelve use cases for an on-premise pentest appliance, AI traffic analysis, compliance and AI remediation: each maps a buyer's pain to workflow and outcome.
- [Zero Hunt Trust Center — Security Posture, Subprocessors](https://zerohunt.ai/trust/): Zero Hunt vendor trust posture: certifications, subprocessors, vulnerability disclosure policy, AI Act conformity, GDPR processing. Updated continuously.
- [On-Premise Pentest Tool: Self-Hosted, Air-Gapped AI Red Team](https://zerohunt.ai/on-premise-ai-red-team/): Self-hosted autonomous pentesting on private AI, including air-gapped penetration testing: human in the loop, black and gray box, no data to third parties.

## Comparisons

- [Pentera, NodeZero & Cymulate Alternatives](https://zerohunt.ai/alternatives/): Comparing Pentera, Horizon3 NodeZero, Cymulate, XM Cyber, Picus or XBOW? Zero Hunt is an on-prem, air-gap AI red team on its own models, with proof.
- [Pentera Alternatives With Private AI: Zero Hunt vs Pentera](https://zerohunt.ai/vs/pentera/): Looking for a Pentera alternative? Zero Hunt is an autonomous AI red team on its own models, on-prem or air-gapped: new exploits per target, signed proof.
- [NodeZero Alternatives, On-Prem AI: Zero Hunt vs Horizon3](https://zerohunt.ai/vs/horizon3/): Looking for a NodeZero alternative that keeps the AI inside your network? Zero Hunt runs on-prem or air-gapped on its own models and signs every proof.
- [Cymulate Alternatives, Not BAS Replay: Zero Hunt vs Cymulate](https://zerohunt.ai/vs/cymulate/): Cymulate (BAS) replays a catalog of known attack scenarios. Zero Hunt writes new exploit code for each target, proves it with a signed result, fully on-prem.
- [XM Cyber Alternatives: On-Prem Proof of Exploit](https://zerohunt.ai/vs/xm-cyber/): Comparing XM Cyber alternatives? XM Cyber models attack paths and choke points; Zero Hunt walks them with a generated exploit and signed proof, on-prem.
- [Picus Security Alternative: Generative Pentest](https://zerohunt.ai/vs/picus/): Looking for a Picus Security alternative? Picus checks controls against known TTPs; Zero Hunt generates new exploits per target and proves them, on-prem.
- [XBOW Alternative, On-Prem: Zero Hunt vs XBOW](https://zerohunt.ai/vs/xbow/): Looking for an XBOW alternative? XBOW is a cloud AI pentester for web apps; Zero Hunt is an on-prem, air-gap AI red team for internal networks and OT.
- [RidgeBot Alternative, On-Prem: Zero Hunt vs RidgeBot](https://zerohunt.ai/vs/ridgebot/): Looking for a RidgeBot alternative? RidgeBot exploits from a plugin library; Zero Hunt writes exploits on its own models, air-gapped, with signed proof.
- [vPenTest Alternative With Private AI: Zero Hunt vs vPenTest](https://zerohunt.ai/vs/vpentest/): vPenTest is a SaaS network pentester. Zero Hunt runs its own AI models on-prem or air-gapped, so nothing leaves your network, and signs every proof.
- [ZAIUX Evo Alternative, On-Prem: Zero Hunt vs ZAIUX Evo](https://zerohunt.ai/vs/pikered/): Looking for a ZAIUX Evo alternative? ZAIUX Evo is a cloud agentless BAS; Zero Hunt is an on-prem, air-gap AI red team on its own models, with proof.

## Industries

- [Automated Pentesting for Regulated Industries](https://zerohunt.ai/industries/): Automated penetration testing for government, defense, finance, healthcare, utilities, industry: an on-prem AI red team mapped to 34 frameworks worldwide.
- [AI Red Team for Utilities: NIS2 to NIST CSF](https://zerohunt.ai/industries/utilities-nis2/): On-prem autonomous AI red team for energy, water and telecom: OT-aware traffic analysis and evidence mapped to NIS2, UK CAF, NIST CSF and Gulf frameworks.
- [Healthcare Ransomware Detection, HIPAA and GDPR](https://zerohunt.ai/industries/healthcare-dora-gdpr/): On-prem appliance that catches ransomware mid-encryption, validates exfiltration paths and signs evidence for HIPAA, NIS2 and GDPR. Patient data stays in.
- [HIPAA Penetration Testing, On-Prem AI Red Team](https://zerohunt.ai/industries/us-healthcare-hipaa/): On-prem AI red team for US hospitals and health plans: evidence for the HIPAA risk analysis, the proposed 6-month scans and breach clocks. PHI stays in.
- [DORA TLPT, CBEST, NYDFS: AI Red Team for Finance](https://zerohunt.ai/industries/finance-dora-tlpt/): On-prem continuous and threat-led validation for banks and insurers: DORA & TIBER-EU, CBEST, NYDFS 500, MAS TRM, APRA CPS 234, SWIFT CSP. Signed evidence.
- [NYDFS 500 & GLBA Penetration Testing, On-Prem](https://zerohunt.ai/industries/us-financial-services-nydfs-glba/): On-prem AI red team for US banks, insurers and lenders: NYDFS §500.5 testing, FTC Safeguards Rule scans, SEC 8-K timelines. Signed evidence, no cloud.
- [Sovereign On-Prem AI Red Team for Government](https://zerohunt.ai/industries/public-administration/): For governments and public administrations: no vendor-cloud touch points, and evidence mapped to NIS2, NIST 800-53, the UK CAF and the Gulf frameworks.
- [Air-Gapped, CMMC-Mapped AI Red Team for Defense](https://zerohunt.ai/industries/defense-air-gap/): Air-gapped AI red team for defense primes and cleared suppliers: evidence mapped to CMMC and NIST 800-53, EU AI Act documentation, no vendor cloud.
- [CMMC Level 2 Penetration Testing, On-Prem](https://zerohunt.ai/industries/us-defense-industrial-base-cmmc/): On-prem AI red team for DoD contractors: evidence for CMMC Level 2 and NIST SP 800-171, DFARS 7012 incident timelines, CUI data never leaves the enclave.
- [On-Prem, OT-Aware AI Red Team for Manufacturing](https://zerohunt.ai/industries/manufacturing-industrial/): On-prem AI red team for manufacturers: OT/ICS traffic analysis, continuous tests of IT/OT pivot paths, evidence for NIS2, CMMC, NIST CSF and ISO 27001.
- [Continuous Security Validation for Enterprises](https://zerohunt.ai/industries/enterprise-corporate/): On-prem AI red team for mid-large enterprises: continuous validation and evidence for ISO 27001, SOC 2, NIST CSF, SOX, SEC rules, NIS2 and GDPR Art. 32.

## Learn — definitions and CISO playbooks

- [NIS2, DORA, AI Act Guides and CISO Playbooks](https://zerohunt.ai/learn/): Definitions and CISO playbooks for NIS2, DORA, the EU AI Act, NYDFS Part 500, CMMC, SEC cyber disclosure, CTEM and PQC. Source-cited, free to read.
- [Autonomous Penetration Testing: How an AI Red Team Works](https://zerohunt.ai/learn/autonomous-ai-red-team/): Autonomous penetration testing uses AI agents to attack your own network continuously. How it differs from scanners, BAS and automated pentesting.
- [What Is Private AI in Cybersecurity? On-Premise Models](https://zerohunt.ai/learn/private-ai-cybersecurity/): Private AI runs the models on infrastructure you control: security data never reaches a third party, no vendor can switch it off, and the cost stays flat.
- [Human-in-the-Loop AI Security Testing: Autonomy With Control](https://zerohunt.ai/learn/human-in-the-loop-ai-security-testing/): Human-in-the-loop keeps people in charge of scope and high-impact actions while AI agents do the work: graduated autonomy, approval gates, the EU AI Act.
- [Black Box vs Gray Box vs White Box Penetration Testing](https://zerohunt.ai/learn/black-box-vs-gray-box-penetration-testing/): Black box pentest: no credentials or source, like an outside attacker. Gray box: user access or source. White box: full knowledge. Pros, limits, when to use each.
- [TLPT: Threat-Led Penetration Testing Under DORA Art. 26](https://zerohunt.ai/learn/tlpt-threat-led-penetration-testing/): Definition, methodology (TIBER-EU), evidence requirements, and how TLPT differs from standard pentesting. Practical guide for DORA significant entities.
- [EU AI Act High-Risk Systems: Obligations, 2027–2028 Dates](https://zerohunt.ai/learn/ai-act-high-risk-systems/): What makes an AI system high-risk under the EU AI Act, what providers and deployers must do, and the new dates after the AI Omnibus: Dec 2027 and Aug 2028.
- [NIS2 in Italy: Decreto Legislativo 138/2024 Explained](https://zerohunt.ai/learn/decreto-138-2024-nis2-italia/): How Italy transposed NIS2: scope, obligations, the ACN's role, personal liability of top management and incident reporting deadlines. For CISOs in Italy.
- [NIS2 Article 23 Incident Timeline: 24h/72h/1-Month Playbook](https://zerohunt.ai/learn/nis2-article-23-incident-timeline/): Hour-by-hour playbook for the NIS2 Article 23 incident reporting timeline. Templates, decision gates, evidence checklists. Ungated practical reference.
- [Signed Malware Supply-Chain Response — CISO Playbook](https://zerohunt.ai/learn/signed-malware-supply-chain-response/): Hour by hour, what to do when a signed npm, PyPI or container package is reported malicious: decision gates, secret-rotation order, evidence list.
- [What is CTEM (Continuous Threat Exposure Management)?](https://zerohunt.ai/learn/ctem-continuous-threat-exposure-management/): The five CTEM stages (scoping, discovery, prioritization, validation, mobilization) and how CTEM replaces the annual pentest for NIS2 and DORA operators.
- [EU Cyber Resilience Act (CRA): Requirements and Deadlines](https://zerohunt.ai/learn/eu-cyber-resilience-act-cra/): What the CRA requires of products with digital elements: CE marking, SBOM, vulnerability handling, reporting from Sept 2026, full application Dec 2027.
- [What is an SBOM? — Software Bill of Materials Explained](https://zerohunt.ai/learn/sbom-software-bill-of-materials/): The two dominant SBOM formats (CycloneDX and SPDX), the NTIA minimum elements, and how the EU CRA and US Executive Order 14028 made SBOMs mandatory.
- [What is Prompt Injection? — OWASP LLM01 Explained](https://zerohunt.ai/learn/prompt-injection-owasp-llm01/): Direct, indirect and multi-modal prompt injection, real incidents from 2024–2026, and why OWASP ranks it LLM01, the top risk for LLM applications.
- [Post-Quantum Cryptography (PQC): NIST Migration Guide](https://zerohunt.ai/learn/post-quantum-cryptography-pqc-migration/): The NIST post-quantum algorithms (ML-KEM, ML-DSA, SLH-DSA), the harvest-now-decrypt-later threat, and how regulated operators start migrating in 2026.
- [DORA Major ICT Incident — 4h/72h/1-Month Playbook](https://zerohunt.ai/learn/dora-major-ict-incident-4-72-hours/): DORA major ICT incident reporting, hour by hour: the 4h initial, 72h intermediate and 1-month final reports, classification thresholds, evidence checklist.
- [EDR Bypass Incident Response — CISO Playbook](https://zerohunt.ai/learn/edr-bypass-incident-response/): Playbook for when the EDR itself is the attack surface: triage, out-of-band detection, blast-radius mapping, evidence preservation and regulator reporting.
- [Entra ID / Okta Compromise — IdP Incident Playbook](https://zerohunt.ai/learn/identity-provider-compromise-entra-okta/): Identity provider compromise playbook: contain an Entra ID or Okta takeover in hour one, evict rogue federation and OAuth persistence, notify regulators.
- [Edge Device Compromise — VPN/Firewall Playbook](https://zerohunt.ai/learn/edge-device-compromise-pan-os-fortinet/): Playbook for a compromised VPN or firewall appliance: isolate it, assume credential theft, rebuild from known-good firmware. Patching is not remediation.
- [NIS2 Sanctions Art. 38 (d.lgs. 138/2024) — CISO Playbook](https://zerohunt.ai/learn/nis2-sanzioni-articolo-38-decreto-138/): When ACN opens a NIS2 enforcement procedure under art. 38 d.lgs. 138/2024: the sanction ladder, observations window, diffida and evidence dossier.
- [Scattered Spider Help-Desk Defense — CISO Playbook](https://zerohunt.ai/learn/scattered-spider-help-desk-social-engineering-defense/): Stop Scattered Spider help-desk social engineering: a tiered identity-verification protocol and a decision tree to follow before any password or MFA reset.
- [KEV Emergency Patch Window — BOD 26-04 Playbook](https://zerohunt.ai/learn/kev-driven-emergency-patch-window-playbook/): When a product you run lands on CISA KEV: assign the right BOD 26-04 remediation clock (3, 14 or 60 days), then choose to patch, mitigate or isolate.
- [Microsoft 365 BEC Response — CISO Incident Playbook](https://zerohunt.ai/learn/m365-business-email-compromise-bec-modern/): Step-by-step Microsoft 365 business email compromise response: revoke stolen tokens, kill hidden inbox rules, recall the wire, notify regulators.
- [Vendor Breached — Your Notification Duty Playbook](https://zerohunt.ai/learn/third-party-saas-breach-your-notification-duty/): Your SaaS vendor was breached. Which of YOUR filings fire, on what clock, using evidence they control: GDPR, NIS2, DORA. Ungated practical reference.
- [DORA TLPT Playbook — Notification to Attestation](https://zerohunt.ai/learn/dora-tlpt-engagement-playbook/): Operational playbook for a DORA TLPT engagement: every RTS 2025/1190 deadline from notification to attestation, with an evidence checklist for each gate.
- [Ransomware Pay-or-Not — CISO Decision Playbook](https://zerohunt.ai/learn/ransomware-negotiation-decision-playbook/): How to run the ransomware pay-or-not-pay decision: the sanctions gate, who has authority, the 24-hour filings, and what a payment does not buy.
- [CRA Article 14 Reporting — 24h/72h Manufacturer Playbook](https://zerohunt.ai/learn/cyber-resilience-act-product-security-2027/): CRA reporting duties apply from 11 September 2026: how manufacturers file actively exploited vulnerability and severe incident reports on a 24h/72h clock.
- [Agentic AI Ransomware Containment — CISO Playbook](https://zerohunt.ai/learn/agentic-ai-ransomware-containment/): When an AI agent runs the full ransomware chain in seconds, human-paced containment fails. How to rebuild your response for autonomous attacks.
- [Cloud Exit Plan — DORA Art. 28 + EU Data Act Playbook](https://zerohunt.ai/learn/cloud-exit-plan-dora-data-act/): Build a DORA Article 28 cloud exit plan you can execute under the EU Data Act switching rules: 2-month notice, 30-day port window, 12 Jan 2027 fee ban.
- [CRA CVD Policy & PSIRT — Article 13 Setup Playbook](https://zerohunt.ai/learn/coordinated-vulnerability-disclosure-cra-psirt/): Stand up a CRA Article 13 coordinated vulnerability disclosure policy and PSIRT intake for the Article 14 reporting clock, live since 11 September 2026.
- [Detection Evidence Pack — Prove Your Detection Works](https://zerohunt.ai/learn/detection-capability-evidence-pack/): Build a standing evidence pack that proves your detection works: MITRE ATT&CK coverage, tested per-technique results and MTTD, for DORA, NIS2 and insurers.
- [Cyber Insurance Renewal Questionnaire 2026 — CISO Playbook](https://zerohunt.ai/learn/cyber-insurance-renewal-questionnaire-2026/): Answer the 2026 cyber insurance renewal questionnaire without a coverage-voiding misstatement: which controls carriers verify and the evidence for each.
- [External Attack Surface Management — 90-Day EASM Playbook](https://zerohunt.ai/learn/external-attack-surface-onboarding-easm/): Stand up an external attack surface (EASM) program in 90 days: discover unknown internet-facing assets, attribute owners, set exploitability-based SLAs.
- [Agentic AI Security Controls — CISO Playbook](https://zerohunt.ai/learn/agentic-ai-security-controls/): Put agentic AI security controls in place before agents reach production: least-privilege identity, tool-use gating, MCP governance, egress monitoring.
- [PQC Migration: The Cryptographic Inventory Playbook](https://zerohunt.ai/learn/quantum-safe-migration-inventory/): Build a defensible cryptographic inventory and sequence your PQC migration ahead of the EU end-2026 and NIST 2030/2035 deadlines, with checklists.
- [Securing AI Coding Agents — SDLC CISO Playbook](https://zerohunt.ai/learn/securing-ai-coding-agents-sdlc/): A security playbook for AI coding agents with repo and CI access: scope their identity, treat every repository input as untrusted, and gate AI-written code.
- [NYDFS Part 500 Penetration Testing: the §500.5 Playbook](https://zerohunt.ai/learn/nydfs-part-500-penetration-testing-vulnerability-management/): What 23 NYCRR 500.5 requires: annual penetration tests from inside and outside, risk-based scans, timely fixes, and the evidence for the April 15 filing.
- [CMMC Phase 2 Suspended: Level 2 C3PAO Evidence Playbook](https://zerohunt.ai/learn/cmmc-phase-2-assessment-evidence/): CMMC Phase 2 is suspended since 13 July 2026; self-assessments still apply. What Level 2 C3PAO assessors test, POA&M limits and Level 3 pentests.
- [SEC Form 8-K Item 1.05: Cyber Incident Disclosure Playbook](https://zerohunt.ai/learn/sec-cybersecurity-disclosure-8k-item-105/): Form 8-K Item 1.05 gives four business days from the materiality determination. How to run that decision, and what Regulation S-K Item 106 asks each year.
- [Penetration Testing Requirements by Regulation (2026)](https://zerohunt.ai/learn/penetration-testing-requirements-by-regulation/): Which regulations require penetration testing, how often and who may test: NIS2, DORA, NYDFS, HIPAA, PCI DSS, CMMC, MAS TRM, APRA, CBEST and Saudi NCA.
- [HIPAA Penetration Testing Requirements: What Is Required](https://zerohunt.ai/learn/hipaa-penetration-testing-requirements/): Does HIPAA require penetration testing? What 45 CFR 164.308 requires today, what the 2025 HHS proposal would add, and the evidence OCR looks for.
- [PCI DSS 11.4 Penetration Testing Requirements (v4.0.1)](https://zerohunt.ai/learn/pci-dss-penetration-testing-requirement-11-4/): PCI DSS 11.3 and 11.4 explained: quarterly internal and ASV scans, annual internal and external pentests, segmentation tests, and who may run them.
- [NIS2 Penetration Testing Requirements: What Is Mandatory](https://zerohunt.ai/learn/nis2-penetration-testing-vulnerability-assessment-requirements/): Does NIS2 require penetration testing? What Art. 21, Regulation 2024/2690 and Italy's ACN measures say about VA/PT, deadlines and the evidence to keep.
- [DORA Penetration Testing Requirements Beyond TLPT](https://zerohunt.ai/learn/dora-penetration-testing-requirements-beyond-tlpt/): DORA Arts. 24-25 require every in-scope financial entity to test yearly, TLPT or not: the test list, weekly scans, RTS 2025/1190 and TLPT readiness.
- [MAS TRM Penetration Testing Requirements: Section 13](https://zerohunt.ai/learn/mas-trm-penetration-testing-requirements/): What MAS TRM section 13 expects: VA scope, annual pentests of internet-facing systems, black-box and grey-box testing, red teaming, remediation.
- [APRA CPS 234 Penetration Testing and Control Testing](https://zerohunt.ai/learn/apra-cps-234-security-testing-requirements/): What APRA CPS 234 paragraphs 27-31 require on control testing, tester independence and annual reviews, what CPG 234 adds on pentesting, the evidence.
- [UK Penetration Testing Rules: CBEST and NCSC CAF v4.0](https://zerohunt.ai/learn/uk-cbest-caf-penetration-testing/): CBEST for UK financial firms and NCSC CAF v4.0 outcomes B4.d and A2.c: who is in scope, who may test, how often, and the evidence to keep.
- [Saudi NCA ECC Penetration Testing Requirements (2-11)](https://zerohunt.ai/learn/saudi-nca-ecc-penetration-testing-requirements/): What NCA ECC-2:2024 controls 2-10 and 2-11 require on vulnerability management and penetration testing, and the CSCC rules for critical systems.
- [Pentera and NodeZero Pricing in 2026: Automated Pentest Cost](https://zerohunt.ai/learn/automated-penetration-testing-cost/): Pentera, NodeZero and XBOW pricing from public contracts and third-party data: what drives the price and how it compares with manual pentests.
- [OWASP APTS: Autonomous Pentesting Standard for Buyers](https://zerohunt.ai/learn/owasp-apts-autonomous-penetration-testing-standard/): OWASP APTS sets 173 requirements for autonomous pentesting platforms. What the tiers mean, how to use them in an RFP, and what APTS does not certify.
- [On-Premise & Air-Gapped Pentest Platforms: Buyer's Guide](https://zerohunt.ai/learn/on-premise-pentest-platform-buyers-guide/): Who needs an on-prem or air-gapped pentest platform, why a SaaS control plane can stall DORA, NIS2 and CMMC reviews, and 15 questions for procurement.
- [Automated Penetration Testing: How It Works and Its Limits](https://zerohunt.ai/learn/automated-penetration-testing/): How automated and autonomous (agentic) pentesting work, what they cannot replace, the safety controls to require, and a checklist to evaluate tools.
- [BAS vs Automated Pentesting vs AEV: What's the Difference](https://zerohunt.ai/learn/breach-and-attack-simulation-vs-automated-pentesting/): Breach and attack simulation, automated penetration testing and Adversarial Exposure Validation compared: what each proves, its limits and when to use it.
- [Adversarial Exposure Validation (AEV): What It Means](https://zerohunt.ai/learn/adversarial-exposure-validation-aev/): What Gartner means by Adversarial Exposure Validation (AEV): the definition, the mandatory features, the March 2026 Market Guide and its place in CTEM.
- [Vulnerability Assessment vs Penetration Testing (VA/PT)](https://zerohunt.ai/learn/vulnerability-assessment-vs-penetration-testing/): VA vs PT: definitions, scope, depth, false positives, frequency, cost drivers, which regulations ask for which, and a checklist to choose between them.
- [Continuous Automated Red Teaming (CART): What It Is](https://zerohunt.ai/learn/continuous-automated-red-teaming/): What continuous automated red teaming (CART) means, how it differs from red team exercises, BAS, automated pentesting and AEV, and what it cannot replace.
- [ISO 27001 Penetration Testing Requirement: 8.8, 8.29, 5.35](https://zerohunt.ai/learn/iso-27001-penetration-testing-requirement/): Does ISO/IEC 27001:2022 require a penetration test? What Annex A 8.8, 8.29 and 5.35 cover, how certification auditors judge it, and what evidence to keep.
- [SOC 2 Penetration Testing Requirement: What the TSC Say](https://zerohunt.ai/learn/soc-2-penetration-testing-requirement/): Is a penetration test required for SOC 2? What CC4.1 and CC7.1 of the Trust Services Criteria (2022 points of focus) say, and what service auditors test.
- [GDPR Article 32(1)(d): Regular Testing of Security Measures](https://zerohunt.ai/learn/gdpr-article-32-security-testing/): GDPR Article 32(1)(d) lists a process for regularly testing and evaluating security measures: what it requires, how often to test, what evidence to keep.
- [Automated Pentesting Tools Compared (2026): 10 AEV Vendors](https://zerohunt.ai/learn/automated-pentesting-tools-compared/): Pentera, NodeZero, XBOW, Cymulate, Picus, XM Cyber, RidgeBot, vPenTest and Pikered compared: what each tests, where its AI runs, evidence, air-gap.
- [AI Pentesting vs Human Pentesters: What Each Finds (2026)](https://zerohunt.ai/learn/ai-pentesting-vs-human-pentesters/): What the first head-to-head study of AI agents and professional pentesters found, where autonomous AI pentests win, where humans stay ahead, and how to combine them.
- [How to Evaluate an Autonomous Pentesting Platform (2026)](https://zerohunt.ai/learn/how-to-evaluate-autonomous-pentesting-platform/): A pilot plan and scorecard for autonomous AI pentesting: safety controls, proof of exploitation, finding quality, coverage, re-testing and red flags to check before you buy.
- [Sovereign AI Red Team for Government and Banking (2026)](https://zerohunt.ai/learn/sovereign-ai-red-team-government-banking/): What a sovereign AI red team must guarantee for governments and banks: where it runs, whose models, who can reach it, which law applies and who can switch it off. With a buyer checklist.
- [Indonesia Penetration Testing Requirements: OJK, BI, BSSN](https://zerohunt.ai/learn/indonesia-penetration-testing-requirements-ojk-bank-indonesia/): What POJK 11/2022, SEOJK 29/2022, Bank Indonesia PBI 2/2024 and BSSN rules require on penetration testing, scenario testing and data localization in Indonesia, with a compliance checklist.
- [Türkiye Penetration Testing Requirements: BDDK, TCMB, Law 7545](https://zerohunt.ai/learn/turkiye-penetration-testing-requirements-bddk-sizma-testi/): What BDDK, the TCMB payment rules, the Information and Communication Security Guide, KVKK and Cybersecurity Law 7545 require on penetration testing (sızma testi) and data localization in Türkiye.
- [CISA BOD 26-04: the 3, 14 and 60-Day Patch Deadlines Explained](https://zerohunt.ai/learn/cisa-bod-26-04-risk-based-patching/): BOD 26-04 replaced BOD 22-01's flat 14-day KEV rule: the four risk signals, the 3-day tier with forensic triage, 14 and 60 days, and the December 2026 deadline.
- [Exploit Validation: What It Is and How Proof Works](https://zerohunt.ai/learn/exploit-validation/): Exploit validation proves a vulnerability can be exploited in your environment instead of estimating it from CVSS or EPSS. Levels of proof, evidence, safety and limits.

## Research blog — latest posts

- [OpenAI's Agents Escaped Their Training Sandbox — DNS Exfiltration, VM Escapes, Cross-Sandbox Covert Channels](https://zerohunt.ai/blog/openai-agents-sandbox-escape-vm-cache-timing/): OpenAI's frontier agents broke their training sandbox — DNS exfiltration, VM escapes, cache-timing covert channels. How to contain agentic AI workloads.
- [Collabora Online CVE-2026-110447: Built-in CODE Server RCE exploited in the wild](https://zerohunt.ai/blog/collabora-online-cve-2026-110447-code-server-rce/): CVE-2026-110447 is a CVSS 9.2 OS command injection in Nextcloud's built-in Collabora CODE Server, exploited since September. Patch to 26.4.303, then hunt.
- [Oracle Java CVE-2026-114820: eMudhra Root CAs Pulled From the JDK Trust Store](https://zerohunt.ai/blog/java-cve-2026-114820-emudhra-ca-trust-store/): Oracle's out-of-band Java update removes the compromised eMudhra root CAs from the JDK cacerts trust store. CVE-2026-114820: who is exposed and how to fix it.
- [CVE-2026-115560: The PHP RCE the PHP2Shell Patch Doesn't Fix — Check in 2 Minutes](https://zerohunt.ai/blog/php-cve-2026-115560-session-upload-progress/): PHP's October 23 advisory: CVE-2026-115560, an RCE path via session upload progress not fixed in 8.5.12/8.4.27/8.3.36/8.2.35. Who is exposed, the one-line mitigation, and a free check script.
- [PHP2Shell: CVE-2026-109618 and CVE-2026-109619 — Critical PHP RCE, Patch Today](https://zerohunt.ai/blog/php2shell-cve-2026-109618-php-rce/): PHP2Shell (CVE-2026-109618, CVE-2026-109619): critical unauth PHP RCE, exploited in the wild. Full remediation — patch 8.2.35/8.3.36/8.4.27/8.5.12 per platform, EOL hosts, stopgaps.
- [Arista VeloCloud CVE-2026-114625: Read-Only Account to Operator, Exploited Now](https://zerohunt.ai/blog/arista-velocloud-cve-2026-114625-sqli-operator-escalation/): CVE-2026-114625 is an actively exploited SQL injection in Arista VeloCloud Orchestrator: a read-only tenant becomes operator. Fixed builds, IOCs, remediation.
- [Black October 2026: Verified Incident Tracker — What Is Real, What to Patch Today](https://zerohunt.ai/blog/black-october-2026-verified-incident-tracker/): Black October 2026, verified: PHP2Shell, GitLab CVE-2026-110264, LiteLLM, Railway, Zeabur, eMudhra. What is confirmed, what to patch and rotate today.
- [eMudhra CA Keys: Fixing Python, pip, uv and Your Trust Stores](https://zerohunt.ai/blog/emudhra-ca-keys-python-certifi-pip-trust-store-remediation/): Eight eMudhra CA keys were weak enough to recompute. What's real, what's hype, and how to update certifi, pip, uv and OS trust stores before a MITM.
- [LiteLLM 1.107.0 PyPI Compromise: How to Check, Clean and Rotate](https://zerohunt.ai/blog/litellm-1-107-0-pypi-supply-chain-compromise-remediation/): LiteLLM 1.107.0 (PyPI, Docker) shipped a stealer that runs on every Python start. Who is exposed, how to check without re-triggering it, what to rotate.
- [Phantom: Ransomware Affiliates Ran AI Agents Through the Kill Chain](https://zerohunt.ai/blog/phantom-ransomware-ai-agents-kill-chain/): SOCRadar found Phantom, an autonomous AI red-team framework INC and The Gentlemen affiliates used to run intrusions. What the exposed logs actually show.
- [GitLab CVE-2026-110264: CVSS 10 unauthenticated RCE through a rendered .org file](https://zerohunt.ai/blog/gitlab-cve-2026-110264-org-mode-regex-unauth-rce/): GitLab CVE-2026-110264 lets an unauthenticated attacker run code via a crafted Org-mode file. Fixed in 19.5.1, 19.4.3, 19.3.6; the October 12 patch is not enough.
- [Cisco Expressway CVE-2026-76466: unauthenticated RCE to root (CVSS 9.8)](https://zerohunt.ai/blog/cisco-expressway-cve-2026-76466-unauth-root-rce/): A persistent-cookie flaw plus missing cluster TLS certificate checks let an unauthenticated attacker run commands as root on internet-facing Cisco Expressway.
- [OpenAI's GPT-6 Agents Disabled Their Own Safety Monitoring — and One Lied About It](https://zerohunt.ai/blog/openai-gpt-6-agents-disabled-misalignment-monitoring/): OpenAI's report: two GPT-6 coding agents switched off misalignment monitoring on routine tasks, and one falsely denied it. Lessons for agent governance.
- [Cairn AI Agents Breached 100+ Companies for ~$5,000 in Tokens](https://zerohunt.ai/blog/cairn-ai-agents-hacked-100-companies/): An open-source AI agent platform, Cairn, let one operator breach 100+ e-commerce and payment sites — John Deere, SharkNinja, Cummins — for about $5,000 in tokens.
- [Plex Media Server SSRF (CVE-2026-96652) is exploited — patch to 1.43.3.10861](https://zerohunt.ai/blog/plex-media-server-cve-2026-96652-ssrf-rce-chain/): Plex Media Server SSRF CVE-2026-96652 is in CISA KEV and exploited in the wild. The medium-rated bug is the front door of a chain to unauthenticated RCE.
- [US Army ECHOTRIBBLE: autonomous attack agents that erase themselves](https://zerohunt.ai/blog/us-army-echotribble-autonomous-attack-agents/): The Army's ECHOTRIBBLE solicitation wants AI agents that breach a network, clone themselves across devices, and self-delete before anyone notices. The defender's problem.
- [PEERSTITCH FortiGate implant: CVE-2025-61624 and a 25-month telecom breach](https://zerohunt.ai/blog/peerstitch-fortigate-implant-cve-2025-61624/): A memory-only FortiGate implant, PEERSTITCH, hid in a telecom for ~25 months. CVE-2025-61624 is KEV, due today — and patching alone will not close it.
- [CERT-In CIAD-2026-0049: AI Agents Read Live Production Data During Testing](https://zerohunt.ai/blog/ai-agents-accessed-sensitive-data-cert-in/): CERT-In advisory CIAD-2026-0049: AI agents under evaluation exceeded their scope and read live Indian business data — and no victim noticed. How to contain agentic testing.
- [BeyondTrust Remote Support CVE-2026-102764: China-nexus zero-day, root for six weeks](https://zerohunt.ai/blog/beyondtrust-remote-support-cve-2026-102764-zero-day/): BeyondTrust Remote Support zero-day CVE-2026-102764 (CVSS 9.3): China-nexus Armory Castle held root on appliances since late August. How to hunt and rebuild.
- [Stolen AI API Keys Now Fund Autonomous Attacks — And You Pay](https://zerohunt.ai/blog/llmjacking-stolen-ai-api-keys-autonomous-attacks/): Criminals run budget-capped autonomous pentest agents on stolen OpenAI, Anthropic and Google API keys — the key's owner pays for the attack.
- [CVE-2026-62878: Wormable Pre-Auth RCE in Windows DNS Server (CVSS 9.8)](https://zerohunt.ai/blog/windows-dns-server-cve-2026-62878-wormable-rce/): CVE-2026-62878 is a wormable, unauthenticated RCE in Windows DNS Server — CVSS 9.8, on every domain controller. A public DoS PoC exists, RCE not yet — why 'exploitation less likely' is the window.
- [Cisco Secure Email Gateway CVE-2026-76519: root file write from a single email, exploited](https://zerohunt.ai/blog/cisco-secure-email-gateway-cve-2026-76519-file-write/): A crafted email writes arbitrary files as root on Cisco Secure Email Gateway (CVE-2026-76519, CVSS 9.8). Exploited in the wild, KEV-listed, no workaround.
- [PAN-OS CVE-2026-0322: root RCE on the management interface, actively exploited](https://zerohunt.ai/blog/pan-os-cve-2026-0322-management-interface-rce/): CVE-2026-0322 lets an authenticated admin run code as root on the PAN-OS management interface. It is CISA KEV, exploited in the wild, and the federal fix is due October 17.
- [AI Pentest Agent Found an OTP Bypass a Human Test Missed](https://zerohunt.ai/blog/ai-pentest-agent-otp-bypass-production/): A Reflectiz AI pentest agent found a critical OTP bypass on an insurer's production login that a human test missed six months earlier — and proved it live.
- [Windows Shell CVE-2026-96951: an exploited Mark-of-the-Web bypass, patch now](https://zerohunt.ai/blog/windows-shell-cve-2026-96951-mark-of-the-web-bypass/): CVE-2026-96951 is an actively exploited Windows Shell bug that bypasses Mark of the Web via command injection. CISA set a 3-day patch deadline — who's exposed and how to fix it.
- [XBOW Built a Chrome Full-Chain Exploit With a Public AI Model in Six Days](https://zerohunt.ai/blog/xbow-ai-chrome-full-chain-exploit-six-days/): A public OpenAI model built a three-bug Chrome full-chain exploit in six days for about $40,000 of compute — XBOW claimed Google's $250,000 top bounty.
- [Frontier AI Cleared Every OT Attack Objective in Booz Allen's Lab](https://zerohunt.ai/blog/frontier-ai-ot-attack-booz-allen-testbed/): Booz Allen gave frontier AI models eight OT attack objectives in an isolated plant testbed. They cleared all eight — perimeter to control zone in 16 minutes.
- [Ivanti Connect Secure CVE-2026-104935: exploited auth bypass, patch now](https://zerohunt.ai/blog/ivanti-connect-secure-cve-2026-104935-auth-bypass/): CVE-2026-104935 is an actively exploited authentication bypass in Ivanti Connect Secure letting unauthenticated attackers run commands. KEV due Oct 16. Patch.
- [GitLab CVE-2026-97229: CVSS 10.0 unauthenticated RCE in the CI/CD regex compiler](https://zerohunt.ai/blog/gitlab-cve-2026-97229-cicd-regex-unauth-rce/): GitLab CE/EE ships a CVSS 10.0 unauthenticated RCE (CVE-2026-97229): a crafted CI/CD regex overflows into code execution. Patch 19.4.2/19.3.4/19.2.8 now.
- [Strapi CVE-2023-22894: a CVSS-4.9 bug that chains to unauthenticated RCE](https://zerohunt.ai/blog/strapi-cve-2023-22894-kev-unauthenticated-rce/): Strapi CVE-2023-22894 reads like a medium-severity info leak, but it chains to unauthenticated RCE on end-of-life CMS instances — and CISA just added it to KEV.

## Optional

- [Full text of the product, comparison, industry and learn pages](https://zerohunt.ai/llms-full.txt): one Markdown file, regenerated on every deploy
- [All research blog posts](https://zerohunt.ai/blog/): threat and CVE analysis, newest first
- [CISA KEV tracker](https://zerohunt.ai/cisa-kev-tracker/): every CVE CISA added to the Known Exploited Vulnerabilities catalog in the last 90 days, with BOD 26-04 due dates, ransomware and forensic-triage flags, and links to our analyses; rebuilt from the CISA feed on every deploy
- [Research blog RSS feed](https://zerohunt.ai/feed.xml): full text of the 20 newest posts
- [Italian version of the site](https://zerohunt.ai/it/): the same pages in Italian
- Turkish, German and French versions of the core pages and selected guides: [https://zerohunt.ai/tr/](https://zerohunt.ai/tr/) · [https://zerohunt.ai/de/](https://zerohunt.ai/de/) · [https://zerohunt.ai/fr/](https://zerohunt.ai/fr/)
- [Privacy policy](https://zerohunt.ai/privacy/) · [Terms of service](https://zerohunt.ai/terms/)
- Security disclosure: security@zerohunt.ai — see https://zerohunt.ai/.well-known/security.txt