Skip to content

v1.36.0

Choose a tag to compare

View all tags
@cport1 cport1 released this 09 Sep 18:19
· 24 commits to main since this release
v1.36.0
d10610b

A detection change on all three servers. No API change. The widget is unchanged apart from its version string.

Changed

Corroborated behavioural evidence blocks sooner. When two independent behavioural categories each reach 0.4, the score is floored at 0.6 and the token is withheld. The bar was 0.5. The DevTools console-attach probe no longer counts as one of the agreeing categories: a developer with the console open contributes that one signal and nothing more.

Measured on the benchmark corpus with the change applied: no human sample reaches two agreeing categories at 0.4, the agent catch rate is unchanged from 0.5, and the false-positive gate passes at 0.00% human FPR.

Affected: integrators risk-banding on the raw score will see sessions that carried two agreeing behavioural categories between 0.4 and 0.5 move to 0.6.

Upgrading

Deploy the server. If you load the widget from the CDN, bump the pinned version and the integrity digest.

Subresource Integrity

fcaptcha.js           sha384-FkasK3b7DsF8ovOYPvprFUabcAnLRXWUi/mNBSRR2+9iwu3wm+rFt5TPDw2M8t/k
dist/fcaptcha.min.js  sha384-z+K4pMTIRLYZuauvHNsdFkzbO0AZe25YeZZv62T/toH1F4LelccITxzNUR9FHzMx
Assets 2
Loading