>_The Grade Distribution0%
    Back to blogARTICLE_READY
    Security8 min readMay 5, 2026

    We Scanned the Moz Top 500. Even the Internet's Giants Are Missing the Basics.

    A full-security scan of 428 reachable domains from the Moz Top 500 reveals that even the most trafficked sites on the internet are missing basic headers, misconfiguring cookies, and leaving open redirects.

    Findings

    Historical Grade Distribution: Launch-Era Snapshot

    Out of 428 reachable Moz Top 500 targets, the historical scan corpus skewed toward B/C outcomes under the launch-era model. The live scanner now uses the stricter canonical scoring system, so this table should be read as a historical distribution rather than current grade eligibility.

    GradeCountShare
    S71.6%
    A6815.9%
    B10624.8%
    C17039.7%
    D276.3%
    F5011.7%

    Key insight

    The pattern is clear: grade correlates directly with how many basic controls are missing. These are not exotic zero-days. They are headers, cookie flags, and redirect validation.

    Threat surface

    The Open Redirect Epidemic

    The single most alarming finding: 235 sites — 54.9% of all reachable sites — have open redirects.

    Google's own properties are among them. play.google.com, docs.google.com, maps.google.com, and multiple regional Google domains all returned 302s to externally-controlled URLs when VibeLeak probed the redirect parameter.

    Why it matters

    An open redirect is not a theoretical concern. It is a phishing enabler. An attacker can craft a link that starts with https://accounts.google.com and ends on a credential-harvesting clone. Users trust the domain. The redirect does the rest.

    TrustScan

    Security Header Gaps: The Easy Wins Everyone Skips

    Missing security headers were the most common category of finding across the entire dataset.

    Referrer-Policy

    285

    66.6% of 428 sites missing this header

    Content-Security-Policy

    229

    53.5% of 428 sites missing this header

    X-Content-Type-Options

    200

    46.7% of 428 sites missing this header

    Strict-Transport-Security

    160

    37.4% of 428 sites missing this header

    X-Frame-Options

    153

    35.7% of 428 sites missing this header

    These are one-line fixes. X-Content-Type-Options: nosniff takes ten seconds to add in nginx, Apache, or Vercel. HSTS is a single header once HTTPS is stable. CSP requires more thought, but a baseline policy is not a week-long project.

    ShieldCheck

    Cookie Misconfiguration: The Silent CSRF and XSS Risk

    Cookie findings were nearly as widespread across the Moz 500 corpus.

    • 188 sites (43.9%) send cookies without SameSite
    • 138 sites (32.2%) send cookies without HttpOnly
    • 108 sites (25.2%) send cookies without Secure

    On HTTPS sites, Secure should be mandatory. HttpOnly on session cookies should be mandatory. SameSite=Lax at minimum should be mandatory. These are not advanced hardening steps. They are the baseline for modern cookie hygiene.

    CloudLeak

    Cloud Exposure: When Public Buckets Leak

    VibeLeak's CloudLeak module found 8 sites with Google Cloud Storage bucket URLs exposed in browser-visible HTML or JavaScript.

    Two of those buckets appeared to be publicly listable, including one on a Google support subdomain where unauthenticated probes returned object listings. Eight out of 428 sounds small — until you consider what a publicly listable bucket can contain. User uploads. Database exports. Backup files. Internal assets. One exposed bucket can undo every other security control on the site.

    Protocol

    The security.txt Desert

    341 sites — 79.7% of the scanned corpus — do not publish a /.well-known/security.txt file with a contact address.

    Low effort, high signal

    Per RFC 9116, security.txt gives researchers a standardized way to report vulnerabilities. Without it, security teams are harder to reach, responsible disclosure takes longer, and vulnerabilities stay open. This is the lowest-effort, highest-signal fix on the entire list. It takes five minutes to publish a text file.

    Spotlight

    Notable Examples: The Good, the Bad, and the Surprising

    • Wikipedia-style reference sites can still carry fixable header gaps such as CSP, X-Frame-Options, and Referrer-Policy.
    • Large search and platform properties frequently showed redirect, cookie, or header findings that would cap them below the top bands today.
    • Amazon regional properties were inconsistent across the historical scan set, which is a useful reminder that security posture is domain-specific.
    • The strict current model reserves 100/100 for zero findings; any site with findings should be discussed by findings and current rescan result, not by stale launch-era grade claims.

    Analysis

    What This Means for Everyone Else

    If the top 500 most popular websites — with their security budgets, dedicated teams, and compliance programs — are averaging 5.4 findings per site, what does that say about the long tail?

    It says the basics are still not basic enough. Most security breaches do not start with a zero-day exploit. They start with a missing header, a cookie without HttpOnly, or a redirect that nobody validated. The Moz 500 scan proves that even sites with infinite resources still ship these gaps. For smaller teams, the risk is higher because the detection time is longer.

    The bottom line

    This is why VibeLeak exists. Not to replace penetration testing, but to make the first pass so cheap and fast that teams run it before every launch. A thirty-second scan that catches an open redirect or a missing CSP is worth more than a quarterly audit that happens after the damage.

    VibeSignal

    A Note on AI Visibility

    VibeLeak also measures AI Signal Score — how discoverable and legible a site is to automated agents, crawlers, and answer engines.

    Across the Moz 500, the top sites generally scored well on AI signals, which makes sense: companies at this scale are already indexed, cited, and routed by AI systems. AI visibility is most valuable for smaller companies and new products trying to get their foot in the door. When you are not already in every training corpus, structured metadata, clean robots policy, and protocol-level breadcrumbs are how agents find you. The Moz 500 scan confirms that discoverability is not the problem for the giants. Security hygiene is.

    Next step

    The Call to Action

    You do not need to be in the Moz Top 500 to be scanned. You just need a public URL.

    Run a VibeLeak scan on your site right now. It takes thirty seconds. You will get a grade, a percentile rank against the live corpus, and a prioritized list of findings with copy-paste remediation steps.

    • If Google and Amazon are still missing Referrer-Policy, there is a good chance your site is too.
    • The difference is that they have teams to catch incidents after they happen. Most teams do not.
    • Scan first. Fix what moves the grade. Recheck to prove it worked.

    Methodology: VibeLeak scanned 428 reachable domains from the Moz Top 500 list using its full public trust surface modules: TrustScan (headers, transport), ShieldCheck (cookies, TLS), ThreatSurface (redirects, auth, rate limiting), CloudLeak (exposed storage), TechScan (version disclosure), DeployCheck (deployment artifacts), and DataLeak (sensitive file exposure). Scans were run on 2026-05-05. This article is a historical corpus snapshot; live grades now use the canonical S 98-100, A 90-97, B 75-89, C 55-74, D 35-54, F 0-34 model, with 100/100 reserved for zero findings.

    Next action

    Run the scanner against your own site

    The article lands hardest when it turns into a fix list. Scan, close the gaps, and recheck.

    Start scan