L1
·
Quiz
·
Lab
L2
·
Quiz
·
Lab
L3
·
Quiz
·
Lab
L4
·
Quiz
·
Lab
Module Test
Module 8 · Lesson 1

The Anatomy of a Recon Report

Structure, hierarchy, and the components that separate actionable intelligence from data dumps.
What makes a recon report persuasive enough to drive a decision?

In 2013, Mandiant published APT1: Exposing One of China's Cyber Espionage Units — a 76-page report that named a specific PLA unit (61398) and documented over 140 victim organizations across 20 industries. What made APT1 land with force was not just the intelligence it contained, but how that intelligence was structured: executive summary, actor profile, TTPs, infrastructure indicators, and appendices. Security leadership could read the first four pages and act. Analysts could mine the appendices for months. The report architecture did the persuasion work.

Why Structure Is Not Optional

Recon findings that lack structure become liability. An analyst who dumps raw WHOIS data, paste-site screenshots, and unformatted Shodan results into a PDF has not written a report — they have archived their browser history. The intelligence value is real; the communication value is near zero.

A properly structured report answers three implicit questions every reader brings: How bad is it? (severity and scope), How do we know? (evidence and methodology), and What do we do? (recommendations). AI tools now make it possible to rapidly scaffold these answers — but the analyst must understand the scaffold before delegating its construction.

The Five-Layer Report Architecture

Professional recon reports share a common architecture regardless of whether they originate from a corporate pen-test team, a threat-intel firm, or a government CERT. The five layers are:

Layer 1 — Executive Summary

One page maximum. Severity rating, scope, business risk, and a single prioritized recommendation. Written for a CISO or VP who will read it in an elevator.

Layer 2 — Scope & Methodology

What was examined, what was excluded, which tools and data sources were used, time window of collection. Establishes credibility and reproducibility.

Layer 3 — Key Findings

Numbered findings in descending severity order. Each finding states: observation, evidence, risk impact, and remediation recommendation.

Layer 4 — Technical Appendices

Raw indicators, full screenshots, tool output logs, WHOIS records, certificate data. Consumed by analysts and engineers — not decision-makers.

Layer 5 — Metadata & Handling

Classification label, TLP designation, author, date, version, distribution list. Critical for legal and operational chain of custody.

The Individual Finding Template

Within Layer 3, each finding follows a repeating template that readers learn to scan efficiently. Mandiant, CrowdStrike, and Rapid7 all use variants of the same structure:

  • Finding ID & Title — Short, scannable identifier (e.g., F-03: Exposed Admin Panel on subdomain.target.com)
  • Severity Rating — Critical / High / Medium / Low / Informational using CVSS or house scale
  • Observation — What was found, in plain language. One to three sentences.
  • Evidence — Screenshot reference, log line, or indicator with timestamp and source tool
  • Risk Narrative — What an adversary could do with this finding; business impact framing
  • Remediation — Specific, actionable, time-bounded recommendation
  • References — CVE, CWE, or OWASP link where applicable
AI Integration Point

AI tools like Claude or GPT-4 can draft finding narratives from structured data — paste a raw Shodan result and request a finding block in the template above. The analyst's job shifts to verification and calibration: confirming the evidence chain, adjusting risk language for the client's sector, and ensuring the remediation is operationally feasible. Speed of drafting doubles; responsibility for accuracy remains entirely with the analyst.

Traffic Light Protocol (TLP) — A Non-Negotiable

Every recon report must carry a TLP designation established by FIRST (Forum of Incident Response and Security Teams). TLP controls sharing scope. Getting this wrong can cause legally significant data leakage or prevent defenders who need information from receiving it.

TLP:RED
  • Not for disclosure; restricted to named recipients only
  • Use for findings that could cause harm if disclosed prematurely
  • No forwarding, no printing outside secure environment
TLP:AMBER
  • Limited disclosure; recipients may share within their org
  • Most common classification for pen-test and recon reports
  • May include AMBER+STRICT limiting to named org only
TLP:GREEN
  • Community disclosure; may share with peer organizations
  • Appropriate for threat-intel feeds and sector sharing
  • Not intended for public posting
TLP:CLEAR
  • No restriction; recipients may share without limitation
  • Appropriate for public threat advisories and CVE notifications
  • Formerly TLP:WHITE prior to 2022 revision
Case Reference

The 2021 HAFNIUM Exchange Server disclosures by Microsoft MSTIC were released TLP:CLEAR, allowing rapid widespread defender action. Contrast with the 2014 Sony Pictures breach, where FBI and US-CERT circulated indicator data under TLP:RED to prevent tipping off the attacker before remediation. The TLP choice was a deliberate tactical decision, not an administrative checkbox.

Key Terms

TLPTraffic Light Protocol — a standardized labeling system for controlling the sharing scope of sensitive information, maintained by FIRST.
Executive SummaryThe first layer of a report; a one-page distillation of severity, scope, and top recommendation designed for non-technical decision-makers.
Finding TemplateA repeating structure (ID, severity, observation, evidence, risk, remediation, references) applied to each discrete vulnerability or exposure.
CVSSCommon Vulnerability Scoring System — a standardized numerical scale (0–10) for rating severity of security vulnerabilities.

Lesson 1 Quiz

The Anatomy of a Recon Report — 4 questions
1. What was the primary reason the 2013 Mandiant APT1 report was operationally effective beyond the intelligence it contained?
Correct. APT1's executive summary, actor profile, TTP section, and indicator appendices served distinct audiences — the structure itself was a force multiplier for the intelligence inside.
Not quite. The report's lasting impact came from its layered architecture enabling different audiences to extract value at different depths of engagement.
2. In the five-layer report architecture, which layer would contain raw Shodan query output and full WHOIS records?
Correct. Raw tool output lives in the appendices, consumed by analysts and engineers rather than decision-makers who read the upper layers.
Incorrect. Raw Shodan output and WHOIS records belong in Layer 4 (Technical Appendices) — they are reference material for analysts, not narrative content for decision-makers.
3. Which TLP designation was deliberately chosen by FBI/US-CERT during the 2014 Sony Pictures response to prevent tipping off the attacker?
Correct. TLP:RED restricts information to named recipients only — the appropriate choice when premature disclosure could compromise an active response operation.
Incorrect. TLP:RED was used — it restricts sharing to named recipients only, preventing the attacker from learning what defenders knew before remediation was complete.
4. When using AI to draft finding narratives from raw tool output, what is the analyst's primary residual responsibility?
Correct. AI accelerates drafting; the analyst owns verification, calibration, and operational feasibility of recommendations. The speed gain is real; the responsibility transfer is illusory.
Not quite. The core residual responsibility is verification of the evidence chain, calibration of risk language to the client's sector, and confirming recommendations are operationally feasible.

Lab 1: Structuring a Finding Block

Practice constructing a properly formatted recon finding with AI assistance

Scenario

You have discovered that a target organization's legacy VPN appliance (Pulse Secure) is running a version vulnerable to CVE-2019-11510 (unauthenticated arbitrary file read). Shodan confirms the device is internet-facing on port 443. Your task is to work with the AI to produce a complete, properly structured finding block using the five-element template from Lesson 1.

Ask the AI to help you draft a finding block for this exposure. Then refine it by asking about severity justification, risk narrative framing, and remediation specificity. Try at least three exchanges — requesting drafts, critiques, and improvements.
OSINT Report Assistant
Lab 1
Ready to help you build a structured finding block. Tell me what you've found and I'll help draft it using the five-element template — or ask me to start with a skeleton and we'll refine from there.
Module 8 · Lesson 2

Writing for Multiple Audiences

The same intelligence, repackaged for the CISO, the engineer, the lawyer, and the board.
How do you communicate the same finding to people with fundamentally different vocabularies and risk tolerances?

When Verizon published the first Data Breach Investigations Report in 2008, it faced a structural problem: the underlying data was forensically granular, but the intended audience was primarily business leadership. The solution was a deliberate bifurcation — a narrative summary with infographics for executives, and a technical appendix with raw incident statistics for practitioners. By the 2015 edition, the DBIR had added a third register: sector-specific callouts for legal, HR, and compliance audiences. The DBIR's longevity owes as much to its audience architecture as to its underlying data quality.

The Four-Audience Model

Every significant recon report has at least four potential audience segments, each processing risk through a different lens. Effective reporting requires deliberately constructing content — sometimes the same content — in multiple registers.

Executive / Board

Risk in business terms: revenue exposure, regulatory fines, reputational damage, competitive intelligence implications. Avoid technical jargon entirely. Use analogies. Quantify where possible.

Security Operations

Indicators, TTPs, attack paths, detection opportunities. Needs enough technical depth to build detections and run incident response runbooks. Wants MITRE ATT&CK mappings.

Engineering / IT

Specific systems, versions, configurations. Needs remediation steps they can actually execute — patch versions, configuration changes, firewall rules. Wants ticket-ready specificity.

Legal / Compliance

Regulatory exposure, notification obligations, evidence preservation, chain of custody. Needs to understand what data was accessible and whether breach-notification thresholds were crossed.

Translating Risk Across Registers

The same finding — an exposed S3 bucket containing customer PII — reads fundamentally differently depending on audience:

Technical Register (SecOps)
  • S3 bucket corp-crm-backup-2022 publicly accessible, no authentication required
  • Contains 847 CSV files, ~2.1M records with name, email, phone, partial payment card data
  • Accessible via unauthenticated GET requests; no S3 server access logging enabled
  • No evidence of prior unauthorized access — access logs absent, CloudTrail not configured
Executive Register (Board)
  • Customer data file containing approximately 2 million records was accessible to anyone on the internet with no password
  • The data includes names, contact information, and partial payment data
  • We cannot determine whether unauthorized parties accessed it before discovery
  • GDPR and CCPA breach-notification obligations may apply; legal review required within 48 hours
AI Rewriting Tool

Large language models excel at register translation. A prompt like "Rewrite this technical finding for a non-technical board audience, expressing risk in business terms and avoiding acronyms" consistently produces usable first drafts. The analyst must verify that business impact claims remain grounded in actual evidence — AI models sometimes escalate or de-escalate risk in translation without explicit justification.

The Risk Quantification Problem

Executives and boards increasingly demand quantified risk — not just "high severity" but dollar figures. The FAIR (Factor Analysis of Information Risk) model, developed by Jack Jones and now maintained by the FAIR Institute, provides a structured probabilistic approach to translating technical findings into financial loss estimates.

AI tools can help apply FAIR-aligned reasoning: given a finding's likely frequency of exploitation and probable loss magnitude (incident response costs, regulatory fines, breach notification), they can produce ranges rather than point estimates. Ranges are more honest and defensible than single figures derived from incomplete data.

In 2018, the Marriott/Starwood breach ultimately cost over $600 million in regulatory fines, litigation, and remediation. The initial recon-phase indicators — legacy Starwood systems running unpatched, exposed management interfaces — were discoverable via passive OSINT. A quantified risk treatment of those indicators at report time would have been materially different from a qualitative "Critical" label.

Precision Over Drama

Resist the temptation to catastrophize findings to secure executive attention. Reports that overstate risk lose credibility when predicted impacts don't materialize. Verizon's DBIR consistently shows that organizations with mature security programs make better decisions from data-grounded, probability-weighted risk statements than from worst-case narratives. Credibility compounds over time.

MITRE ATT&CK Mapping in Reports

For SecOps audiences, mapping recon findings to MITRE ATT&CK provides immediate operational context. Reconnaissance findings map primarily to the Reconnaissance (TA0043) tactic — techniques like Active Scanning (T1595), Search Open Websites/Domains (T1593), and Gather Victim Network Information (T1590) are directly observable via OSINT.

Including ATT&CK technique IDs in the technical register allows SOC analysts to immediately cross-reference existing detections, identify gaps, and update threat models without requiring additional translation from the report author. AI tools can suggest relevant ATT&CK mappings when given a finding description — treat these as starting hypotheses requiring analyst validation against the specific evidence.

FAIR ModelFactor Analysis of Information Risk — a probabilistic framework for translating cybersecurity risk into financial terms using frequency and magnitude variables.
Register TranslationThe deliberate rewriting of technical findings into vocabulary and framing appropriate for a specific audience's knowledge base and decision-making context.
ATT&CK MappingAssociating observed findings with MITRE ATT&CK tactic and technique identifiers to enable immediate operational use by security teams.

Lesson 2 Quiz

Writing for Multiple Audiences — 4 questions
1. The Verizon DBIR's longevity is attributed primarily to which structural decision made in its early editions?
Correct. The DBIR's narrative-plus-appendix architecture allowed it to serve both business leadership and security practitioners simultaneously, which is central to its lasting relevance.
Not quite. The DBIR's durability stems from its deliberate audience bifurcation — executive narrative with practitioner appendices — not from a single technical methodology choice.
2. When translating a finding for a Legal/Compliance audience, what is the most critical element to address?
Correct. Legal and compliance teams need to evaluate regulatory exposure, determine whether breach-notification thresholds were crossed, and assess chain of custody — not technical remediation steps.
Incorrect. Legal/Compliance audiences need regulatory exposure assessment, notification obligation analysis, and clarity on what data was accessible — the technical details are for engineers.
3. What caution does the lesson attach to using AI for register translation of risk findings?
Correct. AI models sometimes shift risk framing during translation — the analyst must verify that business impact claims remain grounded in the actual technical evidence.
Not quite. The lesson flags that AI may escalate or de-escalate risk during translation without flagging the change — the analyst must verify that impact claims remain grounded in evidence.
4. Which MITRE ATT&CK tactic most directly maps to passive OSINT reconnaissance activities?
Correct. TA0043 (Reconnaissance) covers techniques like Active Scanning, Search Open Websites/Domains, and Gather Victim Network Information — directly observable via OSINT methods.
Incorrect. Reconnaissance (TA0043) is the tactic covering OSINT-observable activities including active scanning, searching open websites/domains, and gathering victim network information.

Lab 2: Register Translation Practice

Rewrite a technical finding for four distinct audiences using AI assistance

Scenario

You have confirmed an exposed Elasticsearch database at a target organization containing approximately 4 million customer records with names, email addresses, and plaintext passwords. The database has no authentication. Shodan indexed it 23 days ago. Your technical finding is written — now you need versions for the executive, legal, and engineering audiences.

Ask the AI to help you draft the executive, legal/compliance, and engineering versions of this finding. Then experiment: ask it to add FAIR-style financial risk language to the executive version, or ATT&CK technique IDs for the SecOps version. Aim for at least three exchanges exploring different registers.
OSINT Report Assistant
Lab 2
I can help you translate this finding across audience registers. Share the technical details and tell me which audience version you'd like to start with — executive, legal/compliance, or engineering — and I'll draft it for review.
Module 8 · Lesson 3

Evidence Standards & Chain of Custody

How recon reports survive legal scrutiny — and why most don't.
When does a screenshot become admissible evidence, and when is it just a picture?

In the 2016 FTC action against LabMD — a medical testing company — evidence of a data exposure discovered by the security firm Tiversa became central to a years-long legal dispute. The question was not whether the data was exposed, but how it was discovered, by whom, using what methods, and whether the chain of custody from initial discovery to regulatory submission was intact. The case reached the 11th Circuit Court of Appeals and turned substantially on evidence standards that a recon report had not anticipated needing to meet. The lesson: a finding that cannot prove its own provenance is a finding that cannot be used.

What Chain of Custody Means for OSINT

Chain of custody in traditional digital forensics means documenting every person who handled evidence, every tool that processed it, and every transfer that occurred. In OSINT reporting, the equivalent principle applies to the path from raw discovery to documented finding.

A properly evidenced recon finding must be able to answer: Who collected this data? When exactly (timestamp with timezone)? Using which tool and version? From which source (URL, IP, database)? Was the collection passive (no interaction with target systems) or active? Was the raw artifact preserved and hash-verified?

Timestamping

All evidence must carry UTC timestamps. Screenshot EXIF data, tool log timestamps, and Wayback Machine capture dates should be recorded and cross-referenced where possible.

Hash Verification

Raw artifacts (downloaded files, captured pages) should be SHA-256 hashed immediately upon collection. The hash becomes the integrity anchor — any later modification is detectable.

Tool Documentation

Record tool name, version, and configuration used for each finding. Different versions of Shodan, Maltego, or theHarvester produce different results — version specificity is essential for reproducibility.

Analyst Attribution

Each finding must be attributable to a named analyst. In legal proceedings, anonymous findings carry significantly less weight and may be excluded as hearsay without a sponsoring witness.

Screenshot Standards

Screenshots are the most common — and most problematic — form of evidence in recon reports. An unstructured screenshot proves almost nothing in isolation. A properly captured and documented screenshot can anchor a legal proceeding.

  • Full browser window visible — including address bar (showing full URL), date/time in system clock, browser version if possible
  • Unique request identifier visible — if capturing a web interface, include a query parameter or session ID that confirms the specific request
  • Filename convention — include timestamp in filename (ISO 8601: YYYYMMDD-HHMMSS-UTC) and analyst initials
  • Supplemental log capture — where possible, capture browser console output or tool log alongside screenshot to provide independent corroboration
  • Wayback Machine preservation — submit critical URLs to archive.org immediately upon discovery; the archived capture provides independently timestamped corroboration outside analyst control
  • Hash immediately — compute SHA-256 of the screenshot file before any editing; record in evidence log
AI Evidence Limitation

AI tools cannot generate valid evidence. They can help draft the narrative describing evidence, but any AI-generated summary of a finding must be explicitly labeled as derived from analyst-collected artifacts — not as a primary source itself. Jurisdictions increasingly scrutinize AI-generated text in legal proceedings. Maintain a clear separation between AI-assisted narrative and analyst-verified artifacts.

Passive vs. Active Collection — Legal Significance

The distinction between passive OSINT (observing publicly available information without interacting with target systems) and active recon (probing, scanning, or authenticating against target systems) carries significant legal weight in most jurisdictions.

Under the Computer Fraud and Abuse Act (CFAA) in the United States, accessing a computer system "without authorization" or "exceeding authorized access" can constitute a federal offense. The 2021 Supreme Court ruling in Van Buren v. United States narrowed the CFAA's scope, but the boundary between authorized OSINT and unauthorized access remains contested terrain.

Reports must explicitly state the collection methodology — passive or active — for each finding, and must document any written authorization scope for active collection. A finding derived from active scanning without documented authorization may expose the analyst and their organization to legal liability that eliminates the finding's value entirely.

Documentation Discipline

In 2022, a major consulting firm's penetration test report was partially excluded in subsequent litigation because the engagement letter did not explicitly authorize subdomain enumeration, and the report could not document which findings came from passive OSINT vs. active scanning. The differentiation mattered for liability. Build the habit of tagging every finding with its collection method — it costs nothing and can save everything.

Evidence Log Structure

Every recon engagement should maintain a contemporaneous evidence log — a running record of artifacts collected, separate from the final report. This log becomes the foundation from which the report is drafted and the archive from which individual findings can be re-substantiated if challenged.

Log Entry Fields

Timestamp (UTC) · Analyst · Finding ID · Source (tool + target) · Artifact filename · SHA-256 hash · Collection method (passive/active) · Notes

Storage Requirements

Evidence logs and artifacts must be stored in a write-once or append-only medium where feasible. Cloud storage with versioning enabled (AWS S3 Object Lock, Azure Immutable Blob) provides defensible immutability.

Chain of CustodyDocumentation of every handler, tool, and transfer applied to an evidence artifact from collection through presentation, ensuring integrity and admissibility.
SHA-256A cryptographic hash function producing a 256-bit digest used to verify that a digital artifact has not been altered since collection.
CFAAComputer Fraud and Abuse Act — the primary US federal statute governing unauthorized computer access; relevant to the legal classification of recon collection methods.
Passive OSINTIntelligence collection that does not interact directly with target systems — using search engines, public databases, and cached data rather than probing infrastructure.

Lesson 3 Quiz

Evidence Standards & Chain of Custody — 4 questions
1. What was the core evidentiary problem in the FTC v. LabMD case regarding Tiversa's OSINT discovery?
Correct. The case turned on how the data was discovered, by whom, using what methods, and whether the chain of custody was intact — not just whether the exposure existed.
Not quite. The core problem was that the provenance and chain of custody — who discovered the data, how, and with what authority — could not be adequately established for legal proceedings.
2. What is the recommended immediate action after capturing a screenshot as evidence in a recon report?
Correct. Hashing immediately upon capture, before any editing, creates an integrity anchor. Any later modification is detectable — this is what makes the evidence defensible.
Incorrect. The immediate action is to compute a SHA-256 hash before any editing, creating an integrity anchor that makes the evidence defensible if later challenged.
3. How did the 2021 Van Buren v. United States Supreme Court ruling affect OSINT practitioners?
Correct. Van Buren narrowed CFAA scope but did not resolve the passive/active boundary — the terrain remains contested, making collection method documentation essential.
Incorrect. Van Buren narrowed the CFAA but did not clearly resolve the passive/active boundary for OSINT — the distinction remains legally significant and contested.
4. Why must every finding in a recon report be attributable to a named analyst rather than anonymous?
Correct. In legal proceedings, findings must have a sponsoring witness who can testify to their collection and accuracy — anonymous findings lack this anchor and may be excluded.
Not quite. The legal rationale is that anonymous findings may be excluded as hearsay in proceedings where a sponsoring witness — the named analyst — is required to authenticate the evidence.

Lab 3: Evidence Log Construction

Build a defensible evidence log entry for a complex multi-source finding

Scenario

You have discovered a target organization's exposed Jenkins CI/CD server. The evidence trail involves: a Shodan result (timestamp: 2024-11-14 09:42 UTC), a Wayback Machine capture of the login page (archived 2024-11-10), a screenshot you captured today, and a GitHub repository containing hard-coded credentials that reference the server. Each artifact was collected passively. You need to construct a complete evidence log entry and assess whether any of the collection crossed into active territory.

Ask the AI to help you construct a properly formatted evidence log entry for this multi-source finding. Then probe the passive/active boundary — ask whether accessing the GitHub repo or the Shodan result constitutes "interaction" with the target. Explore at least three angles.
OSINT Report Assistant
Lab 3
Ready to help build your evidence log. Walk me through your artifacts — tool, timestamp, source URL, and how each was collected — and I'll help structure a defensible log entry. We can also explore the passive/active boundary for each source.
Module 8 · Lesson 4

AI-Assisted Report Drafting & Quality Control

Integrating AI into the reporting workflow — where it accelerates, where it fails, and how to verify both.
How do you build a reporting workflow where AI saves hours without introducing errors that cost days?

In 2023, Recorded Future published a methodology note alongside several of their threat intelligence reports disclosing that AI language models had been used to assist with initial drafting, pattern summarization, and indicator clustering. The disclosure was deliberate: the firm wanted clients to understand both the efficiency gains and the editorial layer their analysts applied before publication. This transparency model — AI as accelerant, analyst as gatekeeper — has since become a de facto standard among major threat intelligence vendors, including Mandiant (now part of Google) and Secureworks. The question is no longer whether to use AI in reporting; it is how to build quality controls that make the use defensible.

Where AI Genuinely Helps in Report Drafting

The tasks where AI tools produce the most reliable value in recon reporting are those involving structured transformation — taking data in one form and producing it in another with well-defined criteria. These include:

  • Finding block drafting — convert raw tool output (Shodan JSON, WHOIS text, certificate data) into structured finding blocks following a defined template
  • Register translation — rewrite technical findings in executive, legal, or engineering language; first drafts are typically 80–90% usable
  • Executive summary drafting — synthesize multiple findings into a coherent severity narrative with business impact framing
  • ATT&CK mapping suggestions — given a finding description, suggest relevant technique IDs for analyst validation
  • Remediation recommendation drafting — generate specific patch, configuration, or architecture recommendations from CVE and NVD data
  • Consistency checking — identify inconsistencies in severity ratings, terminology, or cross-references across a long report

Where AI Fails — and Why It Matters

AI language models fail systematically in ways that matter profoundly for security reporting. Understanding these failure modes is not optional — an analyst who cannot identify when AI output has gone wrong will eventually submit a report that damages credibility, triggers legal liability, or enables a bad decision.

Hallucinated CVEs and References

Models confidently cite CVE numbers, CVSS scores, and NVD references that do not exist or apply to different software. Every CVE in an AI-drafted finding must be verified against nvd.nist.gov before publication.

Risk Magnitude Drift

When translating between audiences, AI may escalate a Medium finding to Critical or reduce a High to Low without flagging the change. Severity ratings must be locked before register translation and verified afterward.

Temporal Blindness

AI training cutoffs mean models may cite patch availability, vendor support status, or regulatory requirements that have changed. All time-sensitive claims require independent verification.

False Specificity

AI generates plausible-sounding specific details (IP ranges, file sizes, record counts) that may not match actual evidence. Never use AI-generated specifics without verifying against the original artifact.

The AI-Assisted Reporting Workflow

A defensible AI-integrated reporting workflow separates the AI's contribution from the analyst's verification at each stage. Recorded Future's and Mandiant's published approaches share a common pattern:

Stage 1 — Evidence Collection (Analyst)

All collection is analyst-controlled. AI has no role. Evidence log is maintained contemporaneously. Hash and timestamp all artifacts.

Stage 2 — Structured Data Preparation (Analyst)

Raw artifacts are organized into structured inputs — clean JSON, formatted WHOIS, annotated screenshots. The analyst verifies this input before passing to AI.

Stage 3 — AI Draft Generation

AI drafts finding blocks, register translations, and executive summary from structured inputs. Prompts specify template, audience, and constraints explicitly.

Stage 4 — Analyst Review (Critical Gate)

Every CVE verified. Every severity rating confirmed. Every specific detail cross-checked against original artifact. AI-generated text marked internally for audit trail.

Stage 5 — Peer Review

A second analyst reviews the full report specifically looking for AI failure modes: hallucinated references, severity drift, false specificity, temporal blindness.

Stage 6 — Disclosure Labeling

Report notes AI assistance in methodology section — which sections were AI-drafted and analyst-reviewed. This is increasingly required by client contracts and professional standards.

Prompt Engineering for Report Drafting

The quality of AI-generated finding blocks correlates directly with prompt specificity. A prompt that includes the template structure, audience, severity scale, and explicit instructions to flag uncertainty (e.g., "If you are not certain of a CVE number, write [CVE REQUIRES VERIFICATION] instead of guessing") produces dramatically more usable output than a vague "write a finding about this vulnerability." Invest time in building a reusable prompt library for your reporting workflow.

AI Disclosure Standards — Emerging Norms

The professional community is converging on disclosure norms for AI use in security reports. CREST (Council of Registered Ethical Security Testers), PTES (Penetration Testing Execution Standard), and several national CERTs have begun incorporating AI-use disclosure requirements into their reporting standards.

The current emerging norm is: if AI was used to draft, summarize, or transform content that appears in a final report, the methodology section should state which tools were used, in what capacity, and what analyst review was applied. This is not a legal requirement in most jurisdictions as of 2025, but it is increasingly a contractual requirement in enterprise engagements and a professional ethics expectation.

The underlying rationale is identical to citation standards in academic research: the reader needs to understand the provenance of the claims in order to calibrate their confidence appropriately.

The Accountability Principle

AI tools do not sign reports. Analysts do. Whatever AI contributes to a recon report, the analyst's name on the document represents their personal attestation that the content is accurate, evidence-supported, and appropriately calibrated. The efficiency gain from AI is real; the professional accountability transfer is zero. Sign nothing you have not verified.

AI DisclosureThe emerging professional norm of stating in a report's methodology section which AI tools were used, in what capacity, and what analyst review was applied.
HallucinationAI model behavior where plausible-sounding but false information — such as non-existent CVE numbers — is generated with apparent confidence.
Severity DriftThe AI failure mode where risk ratings shift upward or downward during register translation without explicit justification or analyst detection.
Prompt LibraryA curated collection of reusable, tested prompts optimized for specific reporting tasks — drafting finding blocks, generating executive summaries, or producing ATT&CK mappings.

Lesson 4 Quiz

AI-Assisted Report Drafting & Quality Control — 4 questions
1. What was the significance of Recorded Future's 2023 methodology disclosure about AI use in their threat reports?
Correct. Recorded Future's disclosure pioneered the transparency model that major threat intelligence vendors subsequently adopted — AI accelerates; the analyst edits and takes responsibility.
Incorrect. The disclosure established the transparency model: AI assists with drafting and pattern recognition, but analysts apply an editorial gate before publication. This became industry standard.
2. Which of the following is the most dangerous AI failure mode for a recon report that may enter legal proceedings?
Correct. Hallucinated CVE references — cited as fact in a legal proceeding — can invalidate a finding entirely and expose the reporting analyst to professional and legal consequences.
Not quite. Hallucinated CVE numbers are the most dangerous failure mode — a non-existent or misapplied CVE cited in a legal proceeding can invalidate a finding and expose the analyst to liability.
3. According to the recommended AI-assisted workflow, at what stage should AI tools first contribute to a recon report?
Correct. Evidence collection and structured data preparation are entirely analyst-controlled. AI enters only after the analyst has verified and organized the inputs in Stage 2.
Incorrect. AI enters at Stage 3, after the analyst has collected evidence (Stage 1) and organized it into clean structured inputs (Stage 2). Collection is never AI-delegated.
4. What prompt technique does the lesson recommend to reduce AI hallucination of CVE numbers?
Correct. Instructing the model to flag uncertainty explicitly — rather than guess — converts a dangerous silent failure into a visible placeholder that the analyst knows to verify.
Not quite. The technique is to instruct the AI explicitly to write a placeholder like [CVE REQUIRES VERIFICATION] when uncertain, converting a silent hallucination into a visible flag for analyst review.

Lab 4: Full Report Workflow Simulation

Simulate a complete AI-assisted reporting workflow and stress-test for failure modes

Scenario

You are finalizing a recon report for a financial services client. Three findings have been collected: (1) an exposed RDP port (3389) on an AWS-hosted server with a self-signed certificate and no geo-restriction; (2) an employee LinkedIn profile disclosing internal project codenames and vendor relationships; (3) a misconfigured DNS zone allowing zone transfer from a public resolver. You have clean structured data for each. Your task is to run the AI-assisted workflow: draft finding blocks, generate an executive summary, request ATT&CK mappings, and then deliberately try to make the AI hallucinate — to practice catching those failures.

Work through the workflow: draft one finding block, then ask for an executive summary across all three findings, then ask for ATT&CK mappings. Finally, ask the AI to "include the CVE for the RDP exposure" and observe whether it hallucinates — then discuss how to handle it. Aim for at least four exchanges.
OSINT Report Assistant
Lab 4
Ready to run the full workflow. Tell me which finding you'd like to draft first — RDP exposure, LinkedIn OSINT, or DNS zone transfer — and I'll produce a structured finding block. Then we can move through the executive summary, ATT&CK mapping, and quality control steps.

Module 8 Test

Reporting Recon Findings — 15 questions · 80% required to pass
1. Which layer of the five-layer report architecture is specifically written for a CISO who will read it in an elevator?
Correct. The Executive Summary — maximum one page — is designed for non-technical decision-makers who need severity, scope, and a single top recommendation.
Incorrect. Layer 1 (Executive Summary) is the one-page distillation designed for executive-level readers who need severity, scope, business risk, and one recommendation.
2. In the individual finding template, what is the purpose of the "Risk Narrative" element?
Correct. The Risk Narrative translates a technical observation into adversarial capability and business impact — the "so what" that justifies the severity rating.
Incorrect. The Risk Narrative explains what an adversary could do with the finding and frames the business impact — it is the "so what" section of each finding block.
3. TLP:AMBER+STRICT differs from standard TLP:AMBER in what way?
Correct. AMBER+STRICT adds an additional restriction: the named organization cannot share with peer organizations, only internally within its own structure.
Incorrect. TLP:AMBER+STRICT limits sharing to the specific named organization only — standard AMBER allows sharing within the recipient's peer community.
4. What does the FAIR model provide to security reporting?
Correct. FAIR (Factor Analysis of Information Risk) provides a structured probabilistic approach to expressing risk as probable financial loss ranges — enabling quantified executive communication.
Incorrect. FAIR (Factor Analysis of Information Risk) translates technical risk into financial terms using frequency and loss magnitude variables — enabling quantified board-level communication.
5. The 2018 Marriott/Starwood breach cost over $600 million in total. What is the lesson's OSINT-relevant point about this breach?
Correct. The legacy Starwood systems running unpatched with exposed management interfaces were passively discoverable — a quantified risk report at the time would have been very different from a qualitative Critical label.
Incorrect. The lesson's point is that the pre-breach indicators were discoverable via passive OSINT — illustrating why quantified risk treatment in recon reports matters more than qualitative severity labels.
6. Which MITRE ATT&CK technique specifically covers searching open websites and domains during the reconnaissance phase?
Correct. T1593 (Search Open Websites/Domains) specifically covers using search engines, social media, and other public web resources to gather target information.
Incorrect. T1593 (Search Open Websites/Domains) is the technique covering passive use of search engines and public web resources to gather reconnaissance information.
7. In the FTC v. LabMD case, what was the primary evidentiary deficiency in Tiversa's discovery?
Correct. The case turned on how the data was discovered — by whom, using what methods, with what authority — not just whether the exposure existed. Chain of custody was the decisive issue.
Incorrect. The core deficiency was that chain of custody — provenance, methods, and authorization for the discovery — could not be adequately established for legal proceedings.
8. What does immediately computing a SHA-256 hash of a screenshot achieve in the context of evidence standards?
Correct. The SHA-256 hash is the integrity anchor — if the file changes after collection, the hash will no longer match, making tampering detectable. This is the foundation of forensic evidence integrity.
Incorrect. SHA-256 hashing creates an integrity anchor: if the file is altered after collection, the hash will not match, making modification detectable — essential for forensic defensibility.
9. Which Supreme Court ruling narrowed the CFAA's scope and is relevant to OSINT practitioners' legal exposure?
Correct. Van Buren v. United States (2021) narrowed CFAA's scope regarding "exceeding authorized access" but did not resolve the passive/active boundary for OSINT practitioners.
Incorrect. Van Buren v. United States (2021) narrowed the CFAA's scope — particularly regarding "exceeding authorized access" — but left the passive/active OSINT boundary as contested terrain.
10. When using AI to draft finding blocks, which type of task produces the most reliable output?
Correct. AI excels at structured transformation tasks — taking data in one form and producing it in another with well-defined criteria. These tasks produce the most reliable and usable output.
Incorrect. Structured transformation — converting raw tool output into formatted findings per a defined template — is where AI produces the most reliable output. Prediction and legal judgment are outside its reliable capability.
11. What is "severity drift" in AI-assisted report drafting?
Correct. Severity drift is when AI escalates a Medium finding to Critical or reduces a High to Low during translation — without flagging the change — which can mislead decision-makers.
Incorrect. Severity drift is the AI failure mode where risk ratings change silently during register translation — a Medium becomes Critical or a High becomes Low without the analyst noticing.
12. The 2021 HAFNIUM Exchange Server disclosures were released under TLP:CLEAR. What operational goal did this classification serve?
Correct. TLP:CLEAR allowed the HAFNIUM indicators to propagate globally with no barriers — the explicit goal was speed of defender action, which required unrestricted distribution.
Incorrect. TLP:CLEAR was chosen to maximize distribution speed — enabling every defender globally to act immediately without restriction barriers. The goal was rapid widespread patching.
13. In the recommended AI-assisted reporting workflow, what is the purpose of Stage 5 (Peer Review)?
Correct. Peer review in the AI workflow is specifically targeted at catching the failure modes that the primary analyst may have missed during their own review: hallucinations, drift, and false specifics.
Incorrect. Stage 5 Peer Review is explicitly targeted at AI-specific failure modes — a second set of eyes looking for hallucinated CVEs, severity drift, and plausible-sounding false specifics.
14. What does the lesson recommend regarding AI involvement in the evidence collection stage of a recon report?
Correct. Evidence collection is a no-AI zone in the recommended workflow. All collection decisions, tool selection, and artifact preservation are analyst responsibilities before AI is introduced.
Incorrect. Evidence collection (Stage 1) has no AI involvement — it is entirely analyst-controlled. AI enters only at Stage 3, after the analyst has collected and structured the evidence.
15. Which organization maintains the Traffic Light Protocol standard referenced throughout this module?
Correct. FIRST maintains the TLP standard, which was updated in 2022 — including the renaming of TLP:WHITE to TLP:CLEAR and the addition of TLP:AMBER+STRICT.
Incorrect. FIRST (Forum of Incident Response and Security Teams) maintains TLP, including the 2022 revision that renamed TLP:WHITE to TLP:CLEAR and added TLP:AMBER+STRICT.