#AINewsCollector
Collect, aggregate and sort AI news by popularity.
Core Principles
**Don't just search for "AI news today". ** Pan-search returns SEO aggregation pages and trend prediction articles, and will systematically miss community-level virus spreading phenomena (such as the explosion of open source tools and Meme-level events). A multi-dimensional, hierarchical search strategy must be used.
Workflow
1. Multi-dimensional hierarchical search (at least 8 times, 10-12 times recommended)
Search by the following 6 dimensions in sequence, at least once for each dimension:
Dimension A: Weekly/Newsletter aggregation (highest priority 🔑)
This is the most information dense source, with a single article covering 10+ news items.
Search terms:
- "last week in AI" [current month and year]
- "AI weekly roundup" [current month and year]
- "the batch AI newsletter"
- site:substack.com AI news [current month]
After discovering the weekly newspaper, use web_fetch to get the full text and extract all news clues from it.
Dimension B: Community popularity/viral spread (key dimension 🔑)
Capture bottom-up community hot news, this kind of information is almost impossible to reach through general search.
Search terms:
- "viral AI tool" OR "viral AI agent"
- "AI trending" site:reddit.com OR site:news.ycombinator.com
- "GitHub trending AI" OR "AI open source trending"
- AI buzzing OR "everyone is talking about" AI
- "most popular AI" this week
Dimension C: Product release and model update
Search terms:
- "AI model release" OR "LLM launch" [current month]
- "AI product launch" [current month and year]
- OpenAI OR Anthropic OR Google OR Meta AI announcement
- "Large Model Release" OR "AI New Product"
Dimension D: Financing and Business
Search terms:
- "AI startup funding" [current month and year]
- "AI acquisition" OR "AI IPO"
- "AI Financing" OR "AI Investment"
Dimension E: Research Breakthrough
Search terms:
- "AI breakthrough" OR "AI paper" [current month]
- "state of the art" machine learning
- "AI Paper" OR "Machine Learning Breakthrough"
Dimension F: Regulation and Policy
Search terms:
- "AI regulation" OR "AI policy" [current month and year]
- "AI law" OR "AI governance"
- "AI Regulation" OR "Artificial Intelligence Act"
2. Cross-validation and trapping
After the initial search is completed, check whether there are any omissions:
- If a project/event is mentioned in the Newsletter but not covered by the initial search → Search specifically for the project
- If the same event is mentioned by 3+ different sources → it is most likely a hot topic, search in depth to get more details
- If the hot topics of Chinese media and English media are completely different → cover both sides
3. Search keyword design principles (anti-pattern list)
| ❌ Don’t search like this | ✅ You should search like this | Reasons |
|---|---|---|
| "AI news today February 2026" | "AI weekly roundup February 2026" | The former returns the aggregation page, and the latter returns the planned content |
| "AI news today" | "viral AI tool" + "AI model release" Search separately | Pan-search cannot cover community phenomena |
| "artificial intelligence breaking news" | Search by dimension | Too broad, returning noise |
| Add the specific year, month and day to the search term | Use "this week" "today" "latest" | The date will be biased towards forecast/outlook articles |
| Start writing after only 3 searches | At least 8 times, covering 6 dimensions | 3 searches with less than 30% coverage |
4. Comprehensive judgment of popularity
Evaluate the popularity of each news item (1-5 stars) based on the following signals:
| Signal | Weight | Description |
|---|---|---|
| Multiple media reported the same event | ⭐⭐⭐ High | 3+ sources = confirmed hot spots |
| Evidence of community virus spread | ⭐⭐⭐ High | GitHub star skyrocketed, Twitter refreshed, HN home page |
| From authoritative sources (Dinghui, Dachang official announcement) | ⭐⭐⭐ High | But note that Dachang PR does not mean real hot spots |
| Actual user experience sharing | ⭐⭐ Medium | Someone is actually using it > Just posted it |
| Technological breakthrough/scope of influence | ⭐⭐ Medium | |
| Controversial (safety, ethics discussion) | ⭐⭐ Medium | Controversy often indicates great influence |
| Timeliness (the newer, the hotter) | ⭐ Medium-low | Auxiliary sorting |
5. Output format
Sorted in descending order of popularity, output 15-25 news items:
## 🔥 AI News Express (YYYY-MM-DD)
### ⭐⭐⭐⭐⭐ Most popular
1. **[News Title]**
> One sentence summary (no more than 50 words)
> 🔗 [Source Name](URL)
### ⭐⭐⭐⭐ High Heat
2. ...
### ⭐⭐⭐ Medium heat
...
---
📊 A total of XX news items collected this time | Searched XX times | Coverage dimensions: A/B/C/D/E/F | Update time: HH:MM
6. Deduplication and merging
- When the same incident is reported by multiple reports, merge them into one and select the most authoritative/detailed source.
- Indicate "reported by multiple media" in the abstract to reflect the popularity
- Renamed/renamed projects are considered the same event (such as Clawdbot → Moltbot → OpenClaw)
Recommended news sources
See references/sources.md for details.
Notes
- Prioritize HTTPS links
- If you encounter paywall/inaccessible content, mark "subscription required"
- Remain objective and do not make subjective evaluations of news content
- Do not start output after searching less than 8 times
- If the search result for a certain dimension is empty, change the keyword and search again.