Changelog

Release notes — what we shipped, in order.

Every defense we shipped, the incidents they closed, and the UX changes fitted to the same goal: stop abuse, keep real users fast.

2026-11-15

Lacesse Shield output restructure.

Lacesse Shield replaced the earlier in-browser proof-of-work with an interactive verification surface. Signals that scripts can fake removed from the decision path.

  • Interactive verification. Client renders a challenge; selection is submitted to Lacesse; Lacesse returns a signed verdict.
  • Hidden by default. The "Verify you are human" field appears only after the user tries to submit. Nothing visible for automation to pre-solve against.
  • One verification to rule them all. The previous double-verification path in the signup flow removed; one legitimate check now protects all guarded surfaces.
  • Self-contained payload integrity. Challenge payloads are encrypted and verified on this service. Replays and stale tokens are rejected server-side.
  • Per-environment keying. Shield verification uses an environment-specific secret; no cross-environment leakage.
  • Per-route enforcement. The challenge gates every sensitive action we proxy, not just sign-ups.
  • False-positive review. Confirmed accounts and devices from challenge-fail patterns are reviewed and re-admitted when appropriate.
2026-11-15

Signup surface reduced to one formula.

Removing the adversarial-signup surface leaves three user-facing actions: enter an email, enter a password, choose where you heard about Lacesse. Everything else moved behind challenge-time enforcement.

  • Removed inline captcha from the form.
  • Redirect-to-homepage on missing source removed; error message shown in place.
  • Automatic signup redirects removed; pending-submit states remain visible until the challenge resolves.
  • Auto-login after signup reverted to a single round trip; observe-block-time dropped from 30+ seconds to under two.
2026-11-15

Waves of coordinated automated signups blocked.

Multiple waves of automated account creation attempts against the signup form were detected and stopped. Free-model credit distribution was paused for new accounts while the pattern was characterised.

Accounts curated automatically that match farm profiles are removed within minutes of creation.

2026-11-14

Inbox deliverability gates added before account creation.

Emails failing MX or A-record checks are now rejected prior to account creation. Patterns that previously passed basic syntactic checks but were machine-generated at scale are declined at the frontend.

2026-11-10

Device fingerprinting moved up the funnel.

Device fingerprinting is now evaluated before the account is created, with a sliding rule set updated per wave.

2026-10 to 2026-11

Free-credit distribution rules tightened.

Free-tier promotional credit is no longer assigned to accounts matching detected abuse patterns without additional client-side verification.

The same defenses, ready for any site

Any website can integrate Lacesse Shield.

One script tag plus one secret. Full docs and self-serve signup at /shield/.