Reader pictures and captions
Web and extension Readers now preserve lead and in-article pictures with small captions while filtering author avatars. Media stays separate from grading text.
Product updates
This release simplifies evidence editing, merges adjacent passage removals, keeps overlapping layers independently reversible, and preserves consistent recalculation. The baseline detector rules and scoring weights are unchanged.
Web and extension Readers now preserve lead and in-article pictures with small captions while filtering author avatars. Media stays separate from grading text.
Select a full detection or only part of it. Active and removed text stays grouped as one evidence item, with clear remove and restore actions.
Reports and comparisons can be shared as read-only links that expire with the original History item.
Repeated detector records now share one contextual action, and adjacent removed words stay together as one reversible phrase edit.
July 2026
Highlights remain visible when selected evidence crosses wrapped lines, sentences, or source line breaks.
Deleting an annotation recalculates the report and updates its saved History and Compare results.
Removing or restoring a system detection recalculates the same report and stays synchronized across Reader, History, and Compare.
Signed read-only links follow the original History expiration; single reports begin with Story, while comparisons open directly.
Signed-in web and extension reports save to the same private 30-day report window.
The clean article used for scoring now stays aligned with Reader text and word counts.
Reports place scoring evidence first while keeping supporting publisher context clearly separate.
Public metadata, structured data, keyboard focus, crawl controls, and responsive page behavior were standardized.
Archive
Browse earlier releases by year for a concise record of previous product updates.
Analyze an article and inspect the complete rule-based report.