Source list
The site collects only from the sources below, each listed with its official homepage and feed address. Nothing outside this list ever appears on the site.
35 sources
Vendor education
Anthropic News (Claude for Education)
Anthropic publishes no feed — every conventional RSS path 404s and no page declares one, checked again on 2026-08-19. The sitemap is the honest fallback: 515 URLs, every one carrying a <lastmod> date, 256 of them under /news/. Keyword mode is essential here because most Anthropic news is model releases and safety research, not education. Each candidate costs one page fetch, so maxPerRun is deliberately small.
- Reuse note
- robots.txt says Allow: / and advertises this sitemap. Pages are read to summarize and discarded; the site publishes an excerpt and the link.
- Last checked
- 2026-08-19
ChatGPT for Education Newsletter
Ming's find, and the best first-party education source in the registry — the whole newsletter is about AI in teaching, so relevanceMode is 'always'. Published from Substack under openaiforeducation@substack.com, but openaiforeducation.substack.com 301-redirects to this openai.com subdomain; officialDomains is openai.com so the fetcher follows that redirect. Allowlisting substack.com instead would have opened the fetcher to every publication on the platform.
- Reuse note
- Newsletter body is OpenAI's; this site keeps the feed's own summary and links out. No reuse licence inferred.
- Last checked
- 2026-08-18
Google for Education — Learning & Education blog
The feed the first research pass could not find. It tested /feed and an old Blogger-style path; the working pattern is <category-path>/rss/. Verified: 20 items, channel title 'Learning & Education', on-topic throughout (Gemini practice tests, educator-led AI design). This is the education category feed, so every item qualifies.
- Reuse note
- robots.txt allows the listing path and advertises a sitemap. Feed carries excerpts; the article page supplies the body, which is summarized and discarded.
- Last checked
- 2026-08-19
Khan Academy Blog
Khanmigo and AI-learning product updates mixed with general Khan Academy news, so keyword mode applies.
- Reuse note
- robots.txt sets Crawl-delay: 10; content markup is not a reuse licence.
- Last checked
- 2026-08-18
Microsoft Education Blog
Education-only blog, so every item is on topic. Cadence is roughly weekly to a few per month.
- Reuse note
- robots.txt allows the blog path. WordPress content markup is not a reuse licence.
- Last checked
- 2026-08-18
OpenAI News
Broad company news feed, not an education-only feed: keyword mode filters it down. The openai.com/education landing page returned 403 to the research client; academy.openai.com is used as the homepage instead.
- Reuse note
- robots.txt allows /. Feed ships content:encoded HTML, which is not a reuse licence — this site keeps the summary short and links out.
- Last checked
- 2026-08-18
Government and IGOs
e-Estonia
Estonia's digital-state showcase. Broader than education — four of ten items were education-related and two were about both education and AI — so keyword mode does real narrowing. Kept because it covers the national programmes (AI Leap, the //kood school) that Education Estonia reports from the inside.
- Reuse note
- robots.txt disallows only /wp-admin/. Feed carries full post bodies; the site keeps an excerpt and links out.
- Last checked
- 2026-08-19
Education Estonia
Estonia's English-language education portal, and the answer to a question Finland could not: how a non-English-speaking country's education policy reaches this site at all. Verified 2026-08-19: 10 items, all ten about education, three explicitly about AI, feed carries full text. Estonia's AI Leap programme puts AI tools into every upper-secondary school, which makes this first-party rather than commentary. Keyword mode because the portal covers Estonian education broadly.
- Reuse note
- robots.txt sets an empty Disallow, permitting everything, and advertises a sitemap. The feed carries full post bodies, which is not a reuse licence — the site keeps an excerpt and links out.
- Last checked
- 2026-08-19
European Commission — Education and Training
The education directorate's own feed, found on 2026-08-19 — the first pass looked at policy landing pages and concluded there was none. Ten items, low and irregular volume (one entry was dated 2022), so the date window does real work here. Keyword mode because most items are education policy generally rather than AI.
- Reuse note
- robots.txt allows the content paths. Commission content is generally reusable with attribution; the site still publishes only an excerpt and the link.
- Last checked
- 2026-08-19
GOV.UK — Department for Education
Reinstated on 2026-08-19. The earlier refusal was correct but aimed at the wrong endpoint: the feed found then sat under /search/all*, which robots.txt forbids. This one is the department's own Atom feed and is not covered by any Disallow rule — the full robots.txt was re-read to confirm. Keyword mode is essential: the feed is mostly administrative (data collection guides, funding agreements), and only a small share touches AI.
- Reuse note
- robots.txt disallows /search/all* — this feed is NOT under that path. Most GOV.UK content is Open Government Licence; the site still publishes only an excerpt and the link.
- Last checked
- 2026-08-19
OECD Education and Skills Today
OECD's official education blog, and the way in after oecd.org itself returned 403 to every request. Verified 2026-08-19: 10 items, feed carries full text, genuine education-policy writing rather than press releases. Keyword mode because it covers education broadly, not AI specifically. The oecd.org topic page remains unreachable and is not polled.
- Reuse note
- robots.txt disallows only /wp-admin/. The feed carries full post bodies, which is not a reuse licence — the site keeps an excerpt and links out.
- Last checked
- 2026-08-19
European Commission — Regulatory framework for AI
HomepageNo feed (reading list only)
A policy page rather than a news stream. AI Act obligations reach education providers and edtech vendors, so this belongs on the reading list even without a feed.
- Reuse note
- robots.txt returned 200; no feed and no reuse licence observed.
- Last checked
- 2026-08-18
U.S. Department of Education — AI and the Future of Teaching and Learning
HomepageNo feed (reading list only)
A standing guidance PDF, not a news feed. The department's /ai landing candidates returned 404.
- Reuse note
- Crawler permission is UNVERIFIED: the robots path returned 404 after a redirect.
- Last checked
- 2026-08-18
UNESCO — Artificial intelligence in education
HomepageNo feed (reading list only)
The page returned 403 to the research client, so neither the content nor a feed could be verified. Primary global AI-in-education guidance is consequently absent from the site.
- Reuse note
- robots.txt carries content-signal conditions; scraping needs a permission review first.
- Last checked
- 2026-08-18
教育部全球資訊網 — 即時新聞
HomepageNo feed (reading list only)
The most important gap in this registry. Taiwan's primary school and university AI policy source publishes no RSS, and the site does not do HTML polling, so Taiwanese policy news is currently absent.
- Reuse note
- robots.txt does not disallow this listing; no reuse licence inferred.
- Last checked
- 2026-08-18
數位發展部 — 人工智慧
HomepageNo feed (reading list only)
National AI policy context for Taiwan. No feed found.
- Reuse note
- Crawl permission UNVERIFIED: the conventional robots.txt path returned 404.
- Last checked
- 2026-08-18
Research
arXiv — Computers and Society (cs.CY)
The category RSS exposes one announcement batch, which is incomplete for a weekly collector. The robots-allowed pastweek listing exposes the full weekly announcement set; keyword mode remains mandatory.
- Reuse note
- Abstract and metadata discovery only. Each paper carries its own licence.
- Last checked
- 2026-08-26
arXiv — Human-Computer Interaction (cs.HC)
The category RSS exposes one announcement batch, which is incomplete for a weekly collector. The robots-allowed pastweek listing retains learner, teacher and accessibility announcements across the week.
- Reuse note
- Abstract and metadata discovery only. Each paper carries its own licence.
- Last checked
- 2026-08-26
MIT News — Education
Education topic feed, but not AI-only, so keyword mode applies. A few items per month.
- Reuse note
- Feed excerpts and content markup are not a reuse licence.
- Last checked
- 2026-08-18
arXiv — AI and education query
Inactive: export.arxiv.org/robots.txt currently disallows all paths and arxiv.org/robots.txt explicitly disallows /api. Keep for reference only until automated API access is permitted again.
- Reuse note
- Abstract and metadata discovery only. Each paper carries its own licence.
- Last checked
- 2026-08-26
Computers & Education: Artificial Intelligence
HomepageNo feed (reading list only)
The most directly relevant peer-reviewed journal for this product, and the one it is least allowed to poll. Human reading only.
- Reuse note
- MUST NOT BE AUTOMATED: the RSS host's robots.txt says Disallow: /. The journal page also returned 403 and articles may be paywalled.
- Last checked
- 2026-08-18
MIT Teaching Systems Lab
HomepageNo feed (reading list only)
A valid RSS feed exists but is dormant — the latest ten items are from 2020-03 to 2021-01. Left inactive so a weekly run does not spend a request on a feed that has not moved in five years.
- Reuse note
- robots.txt sets Crawl-delay: 10; no blanket reuse grant.
- Last checked
- 2026-08-18
Stanford HAI News
HomepageNo feed (reading list only)
Tried and withdrawn on 2026-08-19. The sitemap is excellent — 2,032 URLs, all dated, 1,040 under /news — but the article pages are not: four different news pages returned 164KB of HTML containing seven <p> tags, and Readability extracted 58-103 characters from each. The body is not in the server-rendered HTML. A live run confirmed it: 7 of 7 pages unreadable, 0 stories, 7 wasted requests. Reading it would need a headless browser, which is a dependency and a fragility this project does not want. Left inactive rather than spending requests every week for nothing.
- Reuse note
- robots.txt allows / (disallowing only /cms/, /api/, /cp/) and advertises a sitemap. Not currently collected — see notes.
- Last checked
- 2026-08-19
Education media
EdSurge
Covers edtech generally; keyword mode narrows to the AI stories.
- Reuse note
- robots.txt allows public paths, disallows /api/. Summary and link only; no article text is republished.
- Last checked
- 2026-08-18
Education Week
Feed returned items dated in the future relative to the 2026-08-18 verification run. The ingest stage rejects future-dated items, so this cannot push a story into an issue that has not happened yet.
- Reuse note
- robots.txt sets Crawl-delay: 10 and disallows /search. Summary and link only.
- Last checked
- 2026-08-18
Inside Higher Ed
High volume on publishing days; keyword mode is doing most of the work here.
- Reuse note
- robots.txt carries content-signal conditions. Do not republish article text; summary and link only.
- Last checked
- 2026-08-18
Times Higher Education
HomepageNo feed (reading list only)
CDN denied the feed to the research client.
- Reuse note
- RSS candidate returned 403. No reuse licence inferred and articles may be paywalled.
- Last checked
- 2026-08-18
Practitioners
AI4K12 Initiative
Entirely about AI in K-12, so every item qualifies. Publishes sparsely — roughly monthly to quarterly — and will be absent from most issues.
- Reuse note
- robots.txt disallows wp-admin only. Individual resources carry their own licences.
- Last checked
- 2026-08-18
Digital Promise
Implementation and equity-focused district resources; several items per week.
- Reuse note
- robots.txt is empty; no reuse licence inferred. Summary and link only.
- Last checked
- 2026-08-18
Hugging Face Blog
Near-daily and mostly technical; keyword mode keeps only the teaching and AI-literacy posts.
- Reuse note
- robots.txt allows /. Item licences vary by post.
- Last checked
- 2026-08-18
The Batch — DeepLearning.AI
Feed candidates return 404/500, but the sitemap carries 3,045 dated URLs with 3,047 under /the-batch/, most recent 2026-08-14. Mostly general AI rather than education, so keyword mode does the narrowing; the cap is small because each candidate costs a page fetch.
- Reuse note
- robots.txt says Allow: / for everyone and names AI crawlers (including Claude-User) explicitly with Allow: /. Pages are read to summarize and discarded.
- Last checked
- 2026-08-19
1EdTech Consortium
HomepageNo feed (reading list only)
Interoperability and credential standards. No feed.
- Reuse note
- robots.txt returned 200; the tested rss.xml returned 404. Standards carry separate terms.
- Last checked
- 2026-08-18
TeachAI
HomepageNo feed (reading list only)
School-ready AI guidance, and still no feed. Its sitemap has 227 dated URLs, so it COULD be polled — but only 3 sit under /blog and the newest entry is 2026-03-30, five months stale. Not worth a weekly request; revisit if it starts publishing again.
- Reuse note
- robots.txt allows / except API, embedded and SAML paths.
- Last checked
- 2026-08-19
THE Campus
HomepageNo feed (reading list only)
RSS candidate returned 404.
- Reuse note
- Parent-domain robots policy applies; some content is subscription-only.
- Last checked
- 2026-08-18
Taiwan
PanSci 泛科學
The one Chinese-language feed the research pass could verify. Broad science coverage, so keyword mode narrows it.
- Reuse note
- robots.txt allows /. Copyright stays with PanSci and its authors; summary and link only.
- Last checked
- 2026-08-18