Easyling release notes July, 2026

AI traffic has overtaken human traffic over the internet. This has strong implications both for services like Easyling and your website. We focused on both of these alongside smaller improvements of the Dashboard. Read the full article for all the details.

Prevent ingestion of AI content

The JavaScript translation engine can ingest dynamic content. This system can find any content that is displayed in a visitor’s browser and add it to Easyling, ready for translation. The drawback of this is that there can be unwanted content that is added by visitors’ browser extensions. To combat this, we improved the default configuration of new projects to exclude them.

Note that this only applies to new projects going forward. If you wish to add it to your existing project, please add ai_* under Advanced settings → Ignore → By classes.

EU AI Act 50(2)

The EU’s new AI Act means that starting August 2026, AI-created content must be marked in some cases. The exact scenarios for this are beyond the scope of this page. However, if your website uses AI for content generation, and you’re required to mark it, we now offer this capability. This is done via custom attributes. For example data-elpv="ai" means that the given HTML element contains text that isn’t confirmed to be entirely human-translated. This suggests that the translation was at least partially AI edited. There are multiple different markings to show the exact provenance of content. You can read the technical details here.

Choice of preview options

We have multiple technologies for publishing the translations, the translation proxy and the JavaScript translation engine called Crest. For most projects, you quickly decide which one you need. However, sometimes that’s not the case. When that happens, the Dashboard can now offer you previews for both. You can use this to check compatibility with the two options before making a final decision. Note that different, potentially mutually exclusive, configurations may be needed for the two options.

Miscellaneous

We have a few miscellaneous items we changed.

• For our Dropbox integration,
  • We now surface errors that can happen during imports or when the account is linked to the project.
  • It’s now possible to specify custom paths of where the files should go.
  • We fixed access token refreshing when XLIFF export is submitted.
• In the new Workbench,
  • When searching, mid-world hyphens aren’t removed anymore.
  • The live preview now immediately updates when a translation is edited.
  • We fixed a bug that prevented approving or excluding entries.
• Pending segments can now be excluded during export processes.
• ContentGuard’s UI now handles the language settings and AI service integration better.
• Work Packages can now be filtered based on multiple linguists’ contributions.
• Translation statistics are shown for all languages in Explorer.
• We unified the default (empty) appearances of all pages.
• In the language selector, we fixed a bug that accidentally changed unrelated settings when others were changed.
• We fixed a bug where in the crawl list, if a given crawl had a large list of specific pages, only the first 10 would be visible.
• Page titles of the Dashboard 3.0 now match the page that you’re on.
• We fixed a bug in the crawl wizard where after opening the wizard then going to the explorer to change exclusion rules, going back to the crawl wizard wouldn’t reflect the changes.
• When a crawl is set to visit the pages of a sitemap, the actual URL list is now shown in the crawl list.