| <!DOCTYPE html> |
| <html lang="en"> |
| <head> |
| <meta charset="UTF-8" /> |
| <meta name="viewport" content="width=device-width, initial-scale=1.0" /> |
| <title>HIR × OAM Authorized Red-Team Map for LLM Defense</title> |
| <style> |
| :root{ |
| --bg:#08090d; --panel:#111722; --panel2:#171f2b; --ink:#f7efe3; --muted:#b9ad9a; |
| --gold:#e2b44f; --amber:#f3a63a; --red:#e35a5a; --green:#68c77b; --blue:#6bb8ff; |
| --line:rgba(226,180,79,.22); --glass:rgba(255,255,255,.055); |
| } |
| *{box-sizing:border-box} body{margin:0;background:radial-gradient(circle at 15% 0%,rgba(226,180,79,.16),transparent 28rem),radial-gradient(circle at 85% 12%,rgba(227,90,90,.14),transparent 30rem),radial-gradient(circle at 50% 100%,rgba(107,184,255,.10),transparent 34rem),var(--bg);color:var(--ink);font-family:Inter,ui-sans-serif,system-ui,-apple-system,BlinkMacSystemFont,"Segoe UI",sans-serif;line-height:1.55} .wrap{width:min(1180px,calc(100% - 32px));margin:auto}.mono{font-family:ui-monospace,SFMono-Regular,Menlo,Consolas,monospace}.noise{position:fixed;inset:0;pointer-events:none;opacity:.045;background-image:url("data:image/svg+xml,%3Csvg viewBox='0 0 250 250' xmlns='http://www.w3.org/2000/svg'%3E%3Cfilter id='n'%3E%3CfeTurbulence type='fractalNoise' baseFrequency='.9' numOctaves='3' stitchTiles='stitch'/%3E%3C/filter%3E%3Crect width='100%25' height='100%25' filter='url(%23n)'/%3E%3C/svg%3E")} |
| header{padding:70px 0 42px}.kicker{display:inline-flex;gap:10px;align-items:center;border:1px solid var(--line);background:var(--glass);border-radius:999px;padding:8px 12px;color:var(--gold);font-size:12px;text-transform:uppercase;letter-spacing:.16em}.dot{width:8px;height:8px;border-radius:50%;background:var(--amber);box-shadow:0 0 22px var(--amber);animation:p 1.5s infinite}@keyframes p{50%{transform:scale(1.25);opacity:1}0%,100%{opacity:.45}}h1{font-size:clamp(40px,7vw,84px);line-height:.96;letter-spacing:-.055em;margin:22px 0 18px}h1 span{color:var(--gold)}.sub{max-width:860px;color:var(--muted);font-size:clamp(17px,2vw,22px)}.warning{margin-top:24px;display:grid;grid-template-columns:74px 1fr;gap:18px;background:linear-gradient(135deg,rgba(227,90,90,.12),rgba(226,180,79,.055));border:1px solid rgba(227,90,90,.28);border-radius:24px;padding:18px;max-width:920px}.warning .ico{display:grid;place-items:center;border-radius:18px;background:rgba(227,90,90,.12);border:1px solid rgba(227,90,90,.22);font-size:30px;color:var(--red)}.warning p{margin:0;color:var(--muted)}.warning strong{color:var(--ink)} |
| nav{position:sticky;top:0;z-index:10;background:rgba(8,9,13,.8);backdrop-filter:blur(14px);border-bottom:1px solid rgba(226,180,79,.12)}nav .wrap{display:flex;gap:10px;overflow-x:auto;padding:12px 0}nav a{white-space:nowrap;text-decoration:none;color:var(--muted);border:1px solid rgba(226,180,79,.16);border-radius:999px;padding:8px 12px;font-size:13px}nav a:hover{color:var(--ink);border-color:var(--gold)}section{padding:54px 0}.title{display:flex;gap:14px;align-items:center;margin-bottom:22px}.num{color:var(--gold);border:1px solid var(--line);border-radius:12px;padding:8px 10px;background:var(--glass);font-size:12px}h2{font-size:clamp(28px,4vw,44px);margin:0;letter-spacing:-.035em}.lead{max-width:900px;color:var(--muted);font-size:18px}.grid{display:grid;gap:18px}.grid.two{grid-template-columns:repeat(2,1fr)}.grid.three{grid-template-columns:repeat(3,1fr)}.grid.four{grid-template-columns:repeat(4,1fr)}@media(max-width:900px){.grid.two,.grid.three,.grid.four{grid-template-columns:1fr}.warning{grid-template-columns:1fr}} |
| .card{background:linear-gradient(180deg,rgba(255,255,255,.062),rgba(255,255,255,.028));border:1px solid rgba(226,180,79,.16);border-radius:24px;padding:22px;box-shadow:0 18px 70px rgba(0,0,0,.28);position:relative;overflow:hidden}.card:before{content:"";position:absolute;inset:0 0 auto;height:2px;background:linear-gradient(90deg,transparent,var(--gold),transparent);opacity:.34}.card h3{margin:0 0 8px;font-size:20px}.card p{margin:0;color:var(--muted)}.tags{display:flex;flex-wrap:wrap;gap:7px;margin-top:14px}.tag{font-size:11px;text-transform:uppercase;letter-spacing:.08em;border:1px solid rgba(255,255,255,.09);border-radius:999px;padding:5px 8px;color:var(--muted);background:rgba(255,255,255,.055)}.tag.red{color:var(--red);border-color:rgba(227,90,90,.25)}.tag.green{color:var(--green);border-color:rgba(104,199,123,.25)}.tag.blue{color:var(--blue);border-color:rgba(107,184,255,.25)}.tag.gold{color:var(--gold);border-color:rgba(226,180,79,.28)} |
| .matrix{display:grid;grid-template-columns:1fr 1.15fr 1.2fr;gap:1px;background:rgba(226,180,79,.18);border:1px solid rgba(226,180,79,.18);border-radius:22px;overflow:hidden}.cell{background:rgba(17,23,34,.96);padding:16px;min-height:105px}.cell.head{background:rgba(226,180,79,.13);color:var(--gold);font-weight:800;min-height:auto}.cell p{margin:5px 0 0;color:var(--muted);font-size:14px}@media(max-width:820px){.matrix{grid-template-columns:1fr}.cell.head:nth-child(n+2){display:none}} |
| .timeline{position:relative;display:grid;grid-template-columns:repeat(7,1fr);gap:12px;margin-top:24px}.phase{border:1px solid rgba(226,180,79,.18);background:rgba(255,255,255,.045);border-radius:20px;padding:18px 14px;min-height:165px}.phase .n{color:var(--gold);font-size:12px}.phase strong{display:block;margin:8px 0}.phase span{color:var(--muted);font-size:14px}@media(max-width:980px){.timeline{grid-template-columns:1fr}} |
| .switcher{display:flex;gap:10px;flex-wrap:wrap;margin:18px 0}.switcher button{cursor:pointer;border:1px solid rgba(226,180,79,.22);background:rgba(255,255,255,.045);color:var(--muted);padding:10px 12px;border-radius:999px}.switcher button.active{color:var(--ink);border-color:var(--gold);background:rgba(226,180,79,.11)}.scenario{border:1px solid rgba(226,180,79,.18);background:rgba(255,255,255,.04);border-radius:24px;padding:24px}.scenario h3{margin-top:0}.scenario p{color:var(--muted)}.callout{border-left:5px solid var(--gold);background:rgba(226,180,79,.09);border-radius:0 20px 20px 0;padding:18px 22px;color:var(--muted)}.callout strong{color:var(--ink)}footer{border-top:1px solid rgba(226,180,79,.14);padding:44px 0 80px;color:var(--muted)}footer strong{color:var(--ink)} |
| </style> |
| </head> |
| <body> |
| <div class="noise"></div> |
| <header class="wrap"> |
| <div class="kicker"><span class="dot"></span> Authorized Red-Team · Defensive Use Only</div> |
| <h1>Offensive-View Map<br><span>for LLM Defense</span></h1> |
| <p class="sub">A safe adversary-emulation architecture: think like an attacker enough to defend, without publishing attack recipes, bypass instructions, malware logic, credential-theft methods, or operational playbooks.</p> |
| <div class="warning"><div class="ico">⛔</div><p><strong>Hard boundary:</strong> This is not a black-hat manual. It is an authorized red-team / blue-team translation layer. It maps attacker intent, pressure points, and defensive checks at a high level only.</p></div> |
| </header> |
| <nav><div class="wrap"><a href="#premise">Premise</a><a href="#hirbreak">What attackers target</a><a href="#lifecycle">Lifecycle</a><a href="#scenarios">Safe scenarios</a><a href="#rules">Rules of engagement</a><a href="#outputs">Defensive outputs</a></div></nav> |
| <main> |
| <section id="premise" class="wrap"> |
| <div class="title"><span class="num mono">01</span><h2>The Safe Offensive View</h2></div> |
| <p class="lead">A useful defense team must understand the shape of abuse. The goal is not to empower abuse. The goal is to recognize how malicious LLM use degrades truth contact, corrupts execution integrity, and violates human boundaries.</p> |
| <div class="grid three"> |
| <div class="card"><h3>Adversary View</h3><p>What pressure point is the attacker trying to exploit?</p><div class="tags"><span class="tag red">Intent</span><span class="tag red">Pressure</span></div></div> |
| <div class="card"><h3>HIR Failure</h3><p>Which part of Honesty, Integrity, or Respect is being attacked?</p><div class="tags"><span class="tag gold">Diagnosis</span><span class="tag blue">Boundary</span></div></div> |
| <div class="card"><h3>Defensive Control</h3><p>What must the runtime, team, or institution do to contain it?</p><div class="tags"><span class="tag green">Protect</span><span class="tag green">Repair</span></div></div> |
| </div> |
| </section> |
| <section id="hirbreak" class="wrap"> |
| <div class="title"><span class="num mono">02</span><h2>What an Attacker Tries to Break</h2></div> |
| <div class="matrix"> |
| <div class="cell head">Targeted HIR Layer</div><div class="cell head">Offensive Pressure</div><div class="cell head">Defensive Translation</div> |
| <div class="cell"><strong>Honesty</strong><p>Reality contact, source truth, identity truth, evidence boundary.</p></div> |
| <div class="cell"><strong>False context pressure</strong><p>Impersonation, fake urgency, fake authority, synthetic evidence, forged consensus.</p></div> |
| <div class="cell"><strong>Verify reality</strong><p>Identity checks, source provenance, fact separation, uncertainty labels, slower decisions.</p></div> |
| <div class="cell"><strong>Integrity</strong><p>Role fidelity, internal consistency, tool boundaries, audit structure.</p></div> |
| <div class="cell"><strong>Boundary-collapse pressure</strong><p>Tries to turn the system into a contradictory actor that violates its own role.</p></div> |
| <div class="cell"><strong>Preserve structure</strong><p>Scope limits, tool firewalling, policy consistency, logging, reproducible triage.</p></div> |
| <div class="cell"><strong>Respect</strong><p>Dignity, consent, safety, personhood, non-exploitation.</p></div> |
| <div class="cell"><strong>Targeting pressure</strong><p>Coercion, intimidation, grooming, shaming, fraud, manipulation of vulnerable people.</p></div> |
| <div class="cell"><strong>Protect agency</strong><p>Target shielding, escalation, refusal of exploitative assistance, supportive safe redirection.</p></div> |
| </div> |
| </section> |
| <section id="lifecycle" class="wrap"> |
| <div class="title"><span class="num mono">03</span><h2>Authorized Emulation Lifecycle</h2></div> |
| <p class="lead">This is a defensive testing lifecycle. It names phases without providing operational attack steps.</p> |
| <div class="timeline"> |
| <div class="phase"><div class="n mono">01</div><strong>Scope</strong><span>Define authorized system, synthetic targets, allowed tests, and stop conditions.</span></div> |
| <div class="phase"><div class="n mono">02</div><strong>Model Abuse Goal</strong><span>Name the abuse class: fraud, impersonation, data exposure, tool misuse, manipulation, or swarm pressure.</span></div> |
| <div class="phase"><div class="n mono">03</div><strong>Map HIR Break</strong><span>Identify whether the pressure attacks honesty, integrity, respect, or all three.</span></div> |
| <div class="phase"><div class="n mono">04</div><strong>Simulate Safely</strong><span>Use harmless prompts, mock systems, dummy data, and no real targets.</span></div> |
| <div class="phase"><div class="n mono">05</div><strong>Observe Controls</strong><span>Check classification, refusal, transformation, logging, escalation, and recovery behavior.</span></div> |
| <div class="phase"><div class="n mono">06</div><strong>Patch</strong><span>Strengthen runtime gates, policies, memory rules, tool permissions, and review queues.</span></div> |
| <div class="phase"><div class="n mono">07</div><strong>Re-test</strong><span>Confirm improvement without degrading legitimate user help or dignity.</span></div> |
| </div> |
| </section> |
| <section id="scenarios" class="wrap"> |
| <div class="title"><span class="num mono">04</span><h2>Safe Scenario Lens</h2></div> |
| <p class="lead">Click each scenario type. These are abstract test cards for defenders, not attack instructions.</p> |
| <div class="switcher"> |
| <button class="active" data-s="fraud">Fraud pressure</button><button data-s="impersonation">Impersonation</button><button data-s="tool">Tool misuse</button><button data-s="data">Data exposure</button><button data-s="swarm">Automation swarm</button><button data-s="vulnerable">Human targeting</button> |
| </div> |
| <div class="scenario" id="scenario"></div> |
| </section> |
| <section id="rules" class="wrap"> |
| <div class="title"><span class="num mono">05</span><h2>Rules of Engagement</h2></div> |
| <div class="grid two"> |
| <div class="card"><h3>Allowed</h3><p>High-level abuse classification, policy testing, benign red-team prompts, dummy data, synthetic users, mock infrastructure, detection improvement, user-safety training.</p><div class="tags"><span class="tag green">Authorized</span><span class="tag green">Synthetic</span><span class="tag green">Logged</span></div></div> |
| <div class="card"><h3>Not Allowed</h3><p>Real target exploitation, bypass recipes, malware logic, credential theft, phishing kits, evasion methods, private data extraction, instructions that improve abuse capability.</p><div class="tags"><span class="tag red">No Real Targets</span><span class="tag red">No Attack Recipes</span></div></div> |
| <div class="card"><h3>Containment</h3><p>Every test should have scope, owner, time window, dataset boundary, rollback path, review channel, and escalation route.</p><div class="tags"><span class="tag blue">Scope</span><span class="tag blue">Audit</span></div></div> |
| <div class="card"><h3>Respect Constraint</h3><p>Do not use real victims, real private data, or emotional manipulation against uninformed people. Respect is a hard boundary, not branding.</p><div class="tags"><span class="tag gold">Dignity</span><span class="tag gold">Consent</span></div></div> |
| </div> |
| </section> |
| <section id="outputs" class="wrap"> |
| <div class="title"><span class="num mono">06</span><h2>Defensive Outputs</h2></div> |
| <div class="grid three"> |
| <div class="card"><h3>Risk Taxonomy</h3><p>A categorized list of abuse patterns mapped to HIR failures and OAM degradation signals.</p></div> |
| <div class="card"><h3>Control Gaps</h3><p>Where detection, refusal, tool permissions, memory boundaries, or escalation did not hold.</p></div> |
| <div class="card"><h3>Patch Plan</h3><p>Concrete defensive changes: stronger gates, better logging, safer transformations, clearer escalation.</p></div> |
| <div class="card"><h3>User Education</h3><p>Plain-language warnings that help people slow down, verify, and preserve agency under pressure.</p></div> |
| <div class="card"><h3>Runtime Metrics</h3><p>False positives, false negatives, refusal quality, safe-completion quality, escalation latency.</p></div> |
| <div class="card"><h3>Retest Evidence</h3><p>Before/after test notes showing whether controls improved without blocking legitimate use.</p></div> |
| </div> |
| <div class="callout" style="margin-top:24px"><strong>Final invariant:</strong> The only legitimate “offensive” use here is authorized emulation that improves defense. If the map increases abuse capability, it has failed HIR.</div> |
| </section> |
| </main> |
| <footer class="wrap"><strong>HIR × OAM Authorized Red-Team LLM Emulation Map</strong><br>Created and Developed by Collin D. Weber · Defensive adversary-emulation architecture · No operational abuse instructions included</footer> |
| <script> |
| const cards={ |
| fraud:{t:'Fraud pressure',p:'The attacker tries to create urgency, false authority, or false evidence so a person makes a rushed decision.',h:'HIR break: Honesty collapses through false context; Respect collapses when the person becomes a target.',d:'Defense: slow decision flow, verify identity/source, label uncertainty, block transaction-enabling help, route to protective guidance.'}, |
| impersonation:{t:'Impersonation',p:'The attacker tries to borrow another identity, role, institution, or trusted relationship.',h:'HIR break: Honesty fails at identity; Integrity fails when the system accepts a false role.',d:'Defense: provenance checks, role verification, signed channels, cautious output, and refusal to strengthen deception.'}, |
| tool:{t:'Tool misuse',p:'The attacker tries to make the model use tools, code, data, or automation outside legitimate authorization.',h:'HIR break: Integrity fails when execution boundaries are crossed.',d:'Defense: tool firewall, authorization checks, read/write separation, least privilege, sandboxing, and audit logs.'}, |
| data:{t:'Data exposure',p:'The attacker tries to make private, secret, or context-bound information leave its proper boundary.',h:'HIR break: Respect fails through privacy violation; Integrity fails through provenance breach.',d:'Defense: secret detection, access scoping, redaction, memory isolation, and escalation for suspected exposure.'}, |
| swarm:{t:'Automation swarm',p:'The attacker scales low-cost generated variants to overwhelm review, moderation, support, or public attention.',h:'HIR break: Honesty degrades through signal flooding; Integrity degrades through volume pressure.',d:'Defense: rate limits, anomaly detection, provenance tagging, batch review, friction, and coordinated response.'}, |
| vulnerable:{t:'Human targeting',p:'The attacker tries to pressure, shame, groom, extort, isolate, or exploit a vulnerable person.',h:'HIR break: Respect collapses completely; OAM converts a person into an object of manipulation.',d:'Defense: refuse exploitative assistance, provide safety resources, preserve dignity, escalate severe risk, and protect agency.'} |
| }; |
| const box=document.getElementById('scenario'); |
| function render(k){box.innerHTML=`<h3>${cards[k].t}</h3><p><strong>Abstract pressure:</strong> ${cards[k].p}</p><p><strong>${cards[k].h}</strong></p><p><strong>Defensive translation:</strong> ${cards[k].d}</p>`} |
| document.querySelectorAll('.switcher button').forEach(b=>b.addEventListener('click',()=>{document.querySelectorAll('.switcher button').forEach(x=>x.classList.remove('active'));b.classList.add('active');render(b.dataset.s)}));render('fraud'); |
| </script> |
| </body> |
| </html> |
|
|
