News, Media, Press, RSS, Video, Podcast — Source Map
60+ sources. Highest legal exposure of any Vivly category. Read the architectural notes at the bottom before integrating.
CRITICAL LEGAL POSTURE
NYT v. OpenAI (allowed to proceed March 2025, focused on regurgitation of memorized training data) reshapes everything:
- Free/scraped news → store only titles, URLs, snippets, hashes. Never permanent vector store.
- Licensed APIs (Reuters/AP/Bloomberg/NewsAPI.ai) → full text only in transient cache.
- "Ingest" vs "embed for training" must be a hard architectural seam.
- Publishers (NYT, Guardian, Reuters, AP) have begun blocking Wayback Machine and Common Crawl since late 2025.
- Every Vivly answer must cite source + license + retrieval timestamp.
1. News Aggregators
| Source | Pricing | DR | Stability | Notes |
|---|---|---|---|---|
| NewsAPI.org | Free dev 100/day non-commercial; Business $449/mo | 2/5 (no full text) | 4/5 | Cleanest API in category but expensive jump to commercial |
| NewsAPI.ai / Event Registry | ~$60-600+/mo token-based | 5/5 (best enrichment) | 4/5 | Event clustering is the unique value; full text + entities + sentiment |
| GDELT 2.0/3.0 | Free | 5/5 events/entities/tone, 2/5 raw text | 5/5 | 15+ years, 65+ languages, BigQuery export; Google Jigsaw backed. 3.0 rolling out 2026 |
| Bing News Search | — | — | DEAD Aug 11 2025 | Replaced by Azure AI Foundry "Grounding with Bing Search" — Agent-only, 40-483% more expensive |
| Google News (unofficial) | Free RSS / SerpAPI $50-500/mo | 2/5 | 2/5 | RSS endpoints undocumented and can break |
| Webz.io (was Webhose) | $500-5k+/mo | 5/5 | 4/5 | News + blogs + forums + reviews + dark web; archive back to 2008 |
| NewsCatcher | $29-399/mo credit-based | 4/5 | 4/5 | Strong NewsAPI.org alternative |
| MediaStack | Free 500/mo, $9.99-249.99/mo | 3/5 | 3/5 | Owned by apilayer; no full text on any tier |
| NewsData.io | Free 200 credits/day; $199.99/mo Basic | 4/5 | 3/5 | India-strong; full content on paid |
| Currents API | Free 600-1000/day; $30+/mo | 3/5 | 2/5 | Small team; use only for prototype |
| Quantexa (was Aylien) | Enterprise $5K+/mo | 5/5 | 4/5 | Aylien standalone absorbed; indie dev funnel dead |
| ContextualWeb | $0-50/mo via RapidAPI | 2/5 | 1/5 | Orphaned — do not build on this |
2. Press / Wire Services (highest-value, hardest-licensed)
- PR Newswire (Cision) — Free RSS for headlines; enterprise API via Cision One. Press releases are intentionally distributed for republication = low legal risk for free RSS use.
- BusinessWire (Berkshire Hathaway) — Same model. Strong for earnings/8-K coverage.
- GlobeNewswire (Notified) — Best free coverage of European small-caps + biotech.
- Reuters Connect (LSEG) — $25-100k+/yr. Licensing forbids redistribution / AI training without explicit rider. Get the rider in writing.
- AP API — Quote-only. AP signed OpenAI licensing deal in 2023 — that signals (a) licensing exists, (b) standard API access does NOT grant AI training rights.
- Bloomberg B-PIPE / Terminal — $50-200k+/yr; per-seat-licensed. News in B-PIPE is per-seat — don't pipe to multi-user product.
Practical aggregator path: RTPR.io ($99/mo Pro) or Benzinga API — aggregate PRNewswire + BusinessWire + GlobeNewswire + AccessWire in one feed.
3. Blog / RSS Ecosystem
- RSS / Atom (open standard) — Free, universal, 5/5 viability. Discovery via
<link rel="alternate">+ heuristic guess of/feed,/rss,/atom.xml. - Feedly Cloud API — API access requires Enterprise (~$1,600/mo+). Leo AI summaries + market-intel enrichment. Pro/Pro+ user tiers do NOT include API since 2023.
- Inoreader API — Public-app API free with review; commercial custom quote.
- FreshRSS (self-hosted) — Free, full source. Strong fallback if commercial RSS aggregators raise prices.
- NewsBlur — $36/yr Premium or free self-host. Solo-maintainer (Samuel Clay) but 15+ years running.
- Superfeedr (PubSubHubbub / WebSub) — Medium-owned since 2016, uncertain roadmap. Roll your own WebSub hub if core.
4. Newsletter Platforms
- Substack — RSS official (
<sub>.substack.com/feed). Notes only via undocumented JSON (/api/v1/posts) — C&D-prone. Stick to RSS for safety. - Beehiiv API — API included on Scale tier $49/mo+. First-party only.
- Ghost (Content + Admin API) — Free self-host or Ghost(Pro) from $9/mo. Best-documented newsletter CMS API.
5. Video Platforms
- YouTube Data API v3 — Free 10K quota units/day per project.
search.listcosts 100 units (so ~100 searches/day on free). ToS forbids using data for ML training. - YouTube Transcripts (unofficial) — Free libs (
youtube-transcript-api) or managed ($0.001-0.01/transcript via Supadata, SearchAPI). DIY breaks every few months — use managed for production. - Twitch Helix API — Free. 800 points/min for app tokens. Chat (IRC/EventSub) is the richest signal.
- Vimeo API — Free dev tier; AI features (transcript/summary) are Enterprise-only as of 2026.
- Dailymotion API — Free with rate limits. European/French-language video.
- TikTok Research API + Display + Commercial Content — Approval-based. Research requires academic affiliation. Solo devs / commercial AI startups denied by default since late 2023. Practical path: TikAPI/Apify/Bright Data (gray zone).
- Instagram Graph API — Business/Creator accounts only. Instagram Basic Display API died Dec 4, 2024. Personal accounts cannot be programmatically read.
- Facebook Graph API — Post-Cambridge-Analytica gutted. Page management only; not a viable intelligence source.
6. Podcasts
| Source | Pricing | DR | Stability | Notes |
|---|---|---|---|---|
| Listen Notes | Free 300 req/mo; Pro $180/mo (5K) | 5/5 | 4/5 | Best DX; $180/mo is the real entry |
| Podchaser API | Free 25K points/mo | 5/5 | 4/5 | Uniquely strong on people/credit graph |
| iTunes Search API (Apple) | Free | 3/5 (metadata only) | 5/5 | ~20 calls/min/IP. Canonical podcast directory lookup |
| Spotify Podcasts | Free + Premium ($11.99/mo) required for Dev | 3/5 (reduced 2025) | 2/5 | ML training explicitly prohibited in ToS. Audio Features/Analysis removed May 2025 |
| PodcastIndex.org | Free | 5/5 | 4/5 | THE strategic choice for podcast metadata. 4M+ feeds, Podcasting 2.0 transcripts |
| Apple Podcasts directory | Free | 5/5 | 5/5 | iTunesId = universal podcast primary key |
Transcription
- Deepgram Nova-3 — $0.0043/min PAYG (~$0.26/hr). Best DX.
- AssemblyAI Universal-1 batch — $0.12/hr; Universal-2 $0.15/hr. Diarization +$0.02. ~3× cheaper than Deepgram with diarization at scale.
- OpenAI Whisper API — $0.006/min ($0.36/hr), batch only. Groq-hosted Whisper at $0.02/hr cheaper.
7. Long-form / Archives
- Wayback Machine (CDX) — Free, ~60 req/min. Internet Archive under existential legal pressure (record-labels $700M lawsuit, publishers won). 87% drop in news captures May-Oct 2025 as publishers (NYT, Guardian) block archive.org_bot. Don't make this a load-bearing dependency.
- archive.today / archive.ph — Free, no formal API, aggressive bot defenses. Operator anonymous; legal posture ambiguous.
- Common Crawl — Free (S3 egress your cost). ~2-2.3B pages/month. Many publishers (NYT, Reuters, CNN) actively block CC's crawler since 2023. For Vivly: use for public web context, not full-text news caching.
- Internet Archive Scholar — Free. ~25M+ scholarly papers; each carries own license.
8. Region/Language-Specific
- Naver (Korea) — News + Search API, free 25K calls/day per API. Without Naver, you have ~0% Korean coverage (Google has minimal presence).
- Yandex News (Russia) — Editorial content shut down Sept 2022 (transferred to VK/dzen.ru). Use Webz.io or GDELT for Russian-language news.
- Baidu (China) — No clean public news API. Use SerpAPI's Baidu News engine or Apify. Index quality eroded since 2020 — consider Sogou or Toutiao alternatives.
- Indian news — Fragmented. NewsData.io best single source for English-Indian; NewsCatcher + Webz.io for Indic-language depth.
Vivly Prioritization Summary
Must-integrate (free or cheap, high value, low legal risk):
- GDELT 2.0/3.0 — free firehose of global events
- PodcastIndex.org — free, complete podcast graph
- iTunes Search API — canonical podcast ID
- RSS/Atom universal — table-stakes
- Wayback Machine CDX — caveat: rate-limited + legal pressure
- Common Crawl — caveat: news publishers block + AI training optics
Tier-2 (paid but worth it):
- NewsAPI.ai / Event Registry — best enriched news
- Listen Notes + Podchaser — podcast depth
- AssemblyAI — best STT economics
- Beehiiv + Ghost APIs — newsletter first-party
- Feedly Cloud Enterprise — market-intelligence enrichment
Tier-3 (enterprise / vertical):
- Reuters Connect, AP, Bloomberg B-PIPE — finance/news premium
- YouTube Data API — non-negotiable for video
- Twitch Helix — gaming/streaming vertical
Avoid:
- Bing Search API (RETIRED Aug 2025)
- Aylien standalone (absorbed by Quantexa)
- TikTok APIs (gatekept)
- Spotify Web API (locked down 2025)
- Substack JSON scrape (use RSS)
- Instagram Basic Display (DEAD Dec 2024)
- ContextualWeb / Currents (unstable)
Architectural Implications for Vivly
Two-tier ingestion seam. Build "metadata-only" (titles, URLs, summaries) vs "full-text licensed" as separate stores with separate retention.
Robots.txt and AI-training opt-outs. Honor
ai.txt,llms.txt, and User-Agent-specific blocks (GPTBot, ClaudeBot, CCBot) — or get sued.Provenance is your moat. Every Vivly answer cites source + license + retrieval timestamp. This is the architectural answer to NYT v. OpenAI's regurgitation exposure.
The "Bing API died" lesson. Diversify across ≥2 redundant providers per category. GDELT + Webz + NewsAPI.ai is a healthier news stack than any single one.
Wire services are the highest-value, highest-risk slice. PRN/BW/GNW press releases are intended for republication — low risk, high signal. Reuters/AP/Bloomberg are intended for licensed use — treat as contractual integrations.