We know 15k websites are infected. how do we tell them?

A security researcher has identified over 15,000 websites serving malicious JavaScript and is struggling with how to alert their owners without creating a public hit list for attackers or triggering accusations of extortion or spam. Commenters weigh options such as emailing site contacts, using WHOIS and abuse@ addresses, working through hosting providers, CERTs, or browser blocklists like Google Safe Browsing, and even building a “have I been pwned”–style lookup service. The exchange highlights a broader tension between responsible disclosure and rapidly protecting users, as well as systemic problems with third‑party scripts, weak contact practices, and hard‑to-escape blocklists.

Core problem

  • Large set (~15k) of sites are serving malicious JS, often via third‑party “supply chain” scripts.
  • Main question: how to alert site owners at scale without causing new harms (e.g., target lists, permanent blocklisting, accusations of hacking).

Direct notification approaches

  • Traditional method: WHOIS contacts; many say still viable via registrar relay or abuse contacts, despite GDPR and privacy guards.
  • Others suggest:
    • Standard addresses (abuse@, webmaster@, admin@, info@).
    • Scraping contact pages (contact, impressum, etc.).
    • Automating email sending over time to avoid looking like spam.
  • Skepticism:
    • Many addresses dead or unmonitored.
    • Messages can be ignored as scams given high volume of fake “security reports”.

Reporting to intermediaries (hosts, CDNs, CERT, authorities)

  • Proposals to:
    • Notify major hosting providers, CDNs (e.g., via abuse portals), and they cascade to customers or suspend sites.
    • Contact browser vendors / Google Safe Browsing and similar services.
    • Use national CERTs, law enforcement, or EU data‑breach channels; possibly group by country.
  • Concern: can lead to heavy‑handed takedowns of small sites without context.

Publishing a public list

  • One camp: publish all affected domains (similar to “have I been pwned”); owners can self‑check; protects users immediately; silence is seen as unethical.
  • Others warn:
    • It becomes a ready‑made target list for attackers.
    • Once domains land on many blocklists, removal is slow or impossible, devastating for small businesses.
  • Middle‑ground ideas:
    • A “checker” site where owners can prove domain control (DNS/.well‑known) and retrieve details.
    • Limited publication for uncontactable or clearly abandoned sites.

Browser/search blocklists and user protection

  • Many argue user protection is primary: report sites to Safe Browsing et al. so users see red warnings.
  • Counterpoints:
    • Getting unblocked after fixing is inconsistent and painful.
    • But others share experiences where delisting was possible within hours after remediation.

Responsible disclosure vs urgency

  • Some say “responsible disclosure” framing fits poorly: these sites are already actively compromised, not merely vulnerable.
  • Analogy: more like seeing credit‑card skimmers installed; you escalate fast rather than quietly emailing.
  • Yet, there’s sympathy for not instantly nuking responsive, well‑meaning admins.

Structural issues with third‑party JS & web security

  • Strong criticism of third‑party analytics/ads/heatmaps: huge attack surface, weak CSP practices, and opaque, dynamic scripts.
  • Suggestions:
    • Host static, hashed third‑party JS on same origin.
    • Use versioned, immutable scripts; no runtime script injection.
    • Recognize that current practices benefit large ad/analytics ecosystems at cost of speed and safety.

Operational and legal obstacles

  • Mass contact risks:
    • Being flagged as spammer.
    • Misinterpretation as extortion if payment is requested.
    • Potential legal threats for “hacking” even when only observing.
  • Some argue the ethical duty is limited: owners are ultimately responsible; beyond a reasonable notification attempt, do nothing more.

Miscellaneous suggestions and reflections

  • Ideas: crowdsourcing contact discovery; AI/agents to find emails; scanning uploads for malware (old BBS practice); using script‑blocking extensions (NoScript) as end‑user mitigation.
  • Some lament that average users now see malware‑ridden, ad‑heavy experiences as “normal,” making the broader web feel “lost” or irredeemably degraded.