Endpoint: /rest/api/3/search/jql&expand=changelog
HTTP method: GET
JQL query used : jql=“project= AND issuetype=Defect”
It was always like that, at least in June 2025 already.
Most of the content is served by portions, page by page. Endpoint you mentioned doesn’t have paging and changelog might be quite large.
To resolve this conflict Atlassian apparently decided to serve just top 10 changelog entries.
For the real complete changelog there is different API endpoint.
GET /rest/api/3/issue/{issueIdOrKey}/changelog
Or
POST /rest/api/3/changelog/bulkfetch