StudentAayush Nayak
Class6-B
GradeGrade 6
SchoolYBIS
1.
Short Answer
(2 pts)
An email addresses the recipient as the “site owner,” includes a single hyperlink, and no other context. Propose two distinct adversary objectives behind such minimal-content lures and, for each objective, list at least two specific artifacts you would collect to confirm or disprove your hypothesis.
✓ Answer
Model answer: Objectives could include (a) credential harvesting (artifacts: redirection chain and landing page DOM; network traces showing form POST endpoints; TLS certificate SANs and issuer; JavaScript deobfuscation results), and (b) scareware or fake SEO/blacklist notices leading to payment (artifacts: payment iframe origins; merchant IDs; WHOIS/registrar history; passive DNS showing campaign infrastructure reuse). Other valid objectives with concrete artifacts earn credit.
2.
Short Answer
(2 pts)
Design a scoring pipeline that triages URLs in inbound email using lexical, host-based, and behavioral features. Identify at least three features per category and explain how you would calibrate thresholds to minimize both false positives and false negatives.
✓ Answer
Model answer: Lexical: length entropy, homoglyph ratio, suspicious TLDs, presence of @ or excess delimiters; Host-based: domain age, NXDomain burst, hosting AS reputation, wildcard A records; Behavioral: conditional redirects by UA/time/IP, credential POST to cross-origin, drive-by download attempts. Calibration: train/test split on labeled campaigns; ROC/AUC analysis; cost-sensitive thresholds tuned on business impact; periodic drift checks with canary datasets.
3.
Short Answer
(2 pts)
Explain why SPF, DKIM, and DMARC passing results cannot be treated as a definitive safety signal. Provide one concrete attack scenario where all three may pass and propose two compensating controls.
✓ Answer
Model answer: Attackers can send from legitimate services or compromised senders with valid SPF/DKIM and DMARC alignment. Example: abuse of a trusted email/marketing platform to send phishing with authenticated mail. Compensating controls: content and link analysis with detonation; brand lookalike detection; relationship-graph anomaly detection; sender behavioral baselines; URL rewriting with click-time evaluation.
4.
Short Answer
(2 pts)
Outline a safe dynamic analysis workflow for detonating a suspicious link that may conditionally serve payloads only to real users. Include environment preparation, network instrumentation, and methods to bypass basic cloaking.
✓ Answer
Model answer: Use isolated VM with instrumented browser, rotating residential IPs, varied UAs/locales/timezones; capture full PCAP and HAR; enable TLS key logging; simulate user interactions (scroll, click, keypress); replay with different referrers; throttle headless signals; follow all redirects; record DOM mutations/screenshots. To bypass cloaking: multiple geographical vantage points; delayed interactions; multiple visit intervals; cookie/session persistence across visits.
5.
Short Answer
(2 pts)
A link resolves to a login page with branding similar to a major cloud provider, but the POST target differs from the visible origin. Describe the signals indicating credential exfiltration and two immediate containment actions.
✓ Answer
Model answer: Signals: cross-origin form POST to unrelated domain; obfuscated JS collecting additional fields; lack of CSP or CSP allowing exfil domains; mixed content anomalies; TLS cert mismatch with brand; data URI intermediate redirection. Containment: block IOCs at email gateway/DNS/proxy; notify affected users for forced password reset and token revocation; retro-hunt for same infra; submit takedown to host/registrar.
6.
Short Answer
(2 pts)
Compare typosquatting, combosquatting, bitsquatting, and soundsquatting in terms of adversary cost, detection difficulty, and user deception effectiveness. Recommend distinct detection heuristics for each.
✓ Answer
Model answer: Typosquatting: simple edits; high deception; heuristics: Levenshtein distance vs. brand. Combosquatting: brand + legit word; high passability; heuristics: brand token + dictionary collocations. Bitsquatting: bit-flip errors; rare but stealthy at scale; heuristics: Hamming-distance on punycode/ASCII. Soundsquatting: phonetic similarity; heuristics: phonetic encoding (Soundex/Metaphone). Cost/difficulty increases from typo/combo < sound/bit; detection mirrors complexity.
7.
Short Answer
(2 pts)
Given CT logs show a burst of certificates for many subdomains under a newly registered domain, and DNS has wildcard A records with low TTLs, infer likely attacker tactics and infrastructure goals.
✓ Answer
Model answer: Likely domain shadowing or mass subdomain generation for rotation/evasion; wildcarding eases fast-flux-like behavior and tracking evasion; low TTL supports rapid IP changes; CT burst suggests automation or ACME issuance for credibility and TLS padlock lure; goal is resilient phishing/malware hosting across many disposable URLs.
8.
Short Answer
(2 pts)
Propose a regex- and DOM-structure-based detector for HTML credential harvesters that minimizes evasion by template changes. Specify signals that are robust to superficial obfuscation.
✓ Answer
Model answer: Use structural features: presence of two password-like inputs or email+password within same form; labels resembling common identity fields; onsubmit handlers accessing value properties and XHR/fetch to non-origin; hidden inputs with redirection tokens; fingerprinting (navigator, screen) plus form. Regex on action URLs anchored to non-origin with TLD blacklists/entropy thresholds. Normalize by stripping whitespace, decoding HTML entities, resolving relative URLs to reduce obfuscation impact.
9.
Short Answer
(2 pts)
Argue for and against blocking newly registered domains (NRDs) at the email or web proxy for ownership-themed lures. Provide a risk-based policy that balances usability and security.
✓ Answer
Model answer: For: NRDs correlate with abuse; reduces exposure windows. Against: legitimate launches and marketing links break; business friction. Policy: soft block with interstitial on NRDs <30 days unless on allowlist; enable click-time detonation; fast-track allow requests with business justification; continuous monitoring and gradual lift as reputation builds.
10.
Short Answer
(2 pts)
Explain how traffic cloaking based on IP reputation and device fingerprinting undermines sandbox detonation and describe two techniques to recover ground-truth behavior.
✓ Answer
Model answer: Cloakers serve benign content to datacenter IPs/headless browsers, malicious to residential/real fingerprints. Recovery: use residential proxy pools and diverse ASNs; run instrumented real browsers with anti-detection configs; replay with captured cookies/localStorage; time-shift visits; leverage collaborative sensors to aggregate behavior across vantage points.
11.
Short Answer
(2 pts)
An email claims your site will be delisted from search results; the linked page requests payment via cryptocurrency. Identify indicators of a fraud operation across content, payment, and infrastructure layers.
✓ Answer
Model answer: Content: generic threats, urgency, grammar anomalies, lack of verifiable case IDs; Payment: crypto-only, no invoice trail, noncompliant terms, third-party wallets with reuse; Infrastructure: NRD, privacy-protected WHOIS, bulletproof hosting ASNs, recycled templates across passive DNS/Wayback, recent CT certs from free CAs. Cross-layer correlation supports fraud classification.
12.
Short Answer
(2 pts)
You must choose between disabling all links in external emails or applying click-time evaluation. Build a decision framework considering attacker adaptation, user productivity, and residual risk.
✓ Answer
Model answer: Evaluate threat model and business tolerance: full disable reduces risk but impairs workflows; attackers pivot to attachments or QR codes. Click-time offers balance with dynamic analysis, but residual risk from zero-delay payloads remains. Framework: classify users by risk, apply stricter controls to high-risk groups; enable link allowlisting; instrument telemetry; measure incident rates and false positive costs to iterate policy.
13.
Short Answer
(2 pts)
Describe how to use DMARC aggregate (RUA) reports to identify abuse of your brand in ownership-themed phishing, and outline next steps once anomalies are detected.
✓ Answer
Model answer: Parse RUAs for sources sending with your From domain failing alignment; cluster by source IP/AS and volume spikes; correlate subject/link patterns from samples. Next: notify providers; strengthen alignment (SPF include minimization, DKIM key rotation); publish stricter DMARC policy; add BIMI for brand protection; block IOCs; user advisories; legal takedowns where feasible.
14.
Short Answer
(2 pts)
Contrast static WHOIS data with passive DNS and certificate telemetry for adversary infrastructure attribution. When does WHOIS privacy hinder and when can alternative signals compensate?
✓ Answer
Model answer: WHOIS gives registrant/registrar and dates but may be privacy-protected; passive DNS shows resolution history and co-hosted indicators; CT logs expose issuance timing and SAN relationships. When WHOIS is redacted, pivot on NS records, registrar account patterns, hosting AS, cert issuer bursts, subdomain reuse, and analytics IDs to cluster infrastructure despite privacy.
15.
Short Answer
(2 pts)
Devise a controlled experiment to evaluate whether ownership-themed subject lines increase phishing click-through compared to generic lures. Specify experimental units, metrics, and ethical safeguards.
✓ Answer
Model answer: Units: randomized employee cohorts; Treatments: ownership-themed vs. generic lure simulations; Metrics: click-through, credential submission, report rates, time-to-report; Analysis: A/B test with power analysis and confidence intervals; Ethics: informed ongoing training program, non-punitive feedback, opt-outs, data minimization, IRB/security governance approval.
16.
Short Answer
(2 pts)
A redirection chain alternates between HTTP 302 and JavaScript-driven window.location changes across four domains, culminating in a data: URL page. Identify two reliable interception points for detection and justify why they are robust.
✓ Answer
Model answer: Interception at network layer capturing Location headers and referrers (resilient to JS changes), and at DOM mutation observers tracking navigation to data: or blob: URLs (payload-agnostic). Additionally, CSP violation reports for form-action/connect-src can flag exfil attempts regardless of obfuscation.
17.
Short Answer
(2 pts)
Propose a retro-hunt strategy to find historic emails related to a newly discovered phishing domain, using only logs and artifacts (no mailbox content). Include at least three log sources and matching approaches.
✓ Answer
Model answer: Sources: secure email gateway URL rewrite logs; web proxy/DNS resolver queries; CT logs for related SANs; SIEM email metadata (sender IPs, subjects); passive DNS for subdomains. Matching: fuzzy match on registrable domain and edit distance; shared IP/ASN clusters; URL path token patterns; cert fingerprint pivots. Correlate timestamps to identify campaign waves.
18.
Short Answer
(2 pts)
Explain how browser security indicators (padlock, EV/OV info) can create false trust in phishing and detail two UI/technical mitigations enterprises can deploy.
✓ Answer
Model answer: Attackers obtain valid DV certs, showing padlock that users misinterpret as site legitimacy. Mitigations: enterprise browser extensions that highlight trusted internal domains; real-time interstitial warnings for NRDs/lookalikes; deploy DNS filtering with block pages; train users that padlock ≠ legitimacy; instrument Safe Browsing/URL reputation APIs.
19.
Short Answer
(2 pts)
A cloud storage link hosts an HTML file that immediately iframes a phishing page on another domain. Discuss why naive URL allowlists fail and propose layered controls that preserve business use of cloud links.
✓ Answer
Model answer: Allowlists trust top-level domains, but embedded iframes redirect to malicious origins, bypassing URL checks. Controls: content inspection of hosted HTML; CSP enforcement on enterprise browsers; sandbox detonation with DOM analysis; tenant allowlisting (restrict to org-owned cloud tenants); block third-party iframes; use cloud provider DLP and app control (CASB).
20.
Short Answer
(2 pts)
Draft a playbook for responding to a widespread campaign targeting site administrators with ownership-themed lures. Include detection, containment, eradication, and recovery steps, and define measurable exit criteria.
✓ Answer
Model answer: Detection: IOC ingestion, retro-hunt, user reports triage; Containment: block domains/IPs, quarantine emails, takedown requests; Eradication: reset creds, revoke tokens, wipe infected endpoints, remove persistence; Recovery: monitor for reoccurrence, restore affected accounts, update detections and training. Exit criteria: zero new hits over two campaign cycles; no active IOCs in telemetry; completion of user remediation and post-incident review.
21.
Fill in the Blank
(1 pts)
A URL that decodes from a punycode form beginning with xn-- often indicates a potential _____ attack against users.
✓ Answer
IDN homograph attack
22.
Fill in the Blank
(1 pts)
A DMARC policy that requests mailbox providers to quarantine messages failing alignment is set with p=_____.
23.
Fill in the Blank
(1 pts)
For DMARC SPF alignment, the RFC5321 envelope domain (also called the _____ domain) is compared to the RFC5322.From domain.
24.
Fill in the Blank
(1 pts)
Serving benign content to sandboxes while delivering payloads to real users based on IP or fingerprint is known as _____.
✓ Answer
traffic cloaking
25.
Fill in the Blank
(1 pts)
Using many randomized subdomains under a single registrable domain to evade reputation systems is characteristic of _____.
✓ Answer
domain shadowing
26.
Fill in the Blank
(1 pts)
Receivers annotate authentication outcomes in the email header field named _____.
✓ Answer
Authentication-Results
27.
Fill in the Blank
(1 pts)
HTML forms can exfiltrate data to a different origin without needing CORS because browsers permit _____.
✓ Answer
cross-origin form submission
28.
Fill in the Blank
(1 pts)
Detecting lookalike domains by minimal edit operations commonly relies on the _____ distance algorithm.
29.
Fill in the Blank
(1 pts)
A DNS configuration that maps any undefined subdomain to a single IP is implemented via a _____ record.
30.
Fill in the Blank
(1 pts)
Proactively discovering brand abuse via newly issued TLS certificates is enabled by monitoring _____ logs.
✓ Answer
certificate transparency