CyberMax-tools/superintelligence-ban-bill-coverage

Policywren: the superintelligence ban bill in news coverage and search interest (Sept 2026)

<!-- cm:banner --> <!-- /cm:banner -->

<!-- cybermax-bridge:start -->

Need it fresh, filtered or via API? This free file is a snapshot (coverage at the last refresh), last updated 2026-09-25. - [Swellmeter on Apify](https://data.cybermax-tools.workers.dev/buy/swellmeter?s=hf-superintelligence-ban-bill-coverage) ($0.003 per keyword analyzed): checks Google Trends direction for your own keywords with a plain rising/falling verdict, daily if scheduled. - [Get an email when this dataset updates](https://data.cybermax-tools.workers.dev/notify?ds=superintelligence-ban-bill-coverage&s=hf-superintelligence-ban-bill-coverage): free, double opt-in, unsubscribe any time.

<!-- cybermax-bridge:end -->

In September 2026 "superintelligence" became a US policy topic: a US senator and a House member announced a bill to ban the development of artificial superintelligence and pause advanced AI development until a federal regulator sets safety rules (Senate press release, 3 Sep; widely reported on 23 Sep), and a White House release quoted a 22 Sep UN speech saying AI is "hereinafter officially called 'Super Intelligence'". This dataset shows how much attention that got, day by day, from public sources, with the primary sources re-checked on the day it was built.

Three tables (built 24 Sep 2026, 22:40 UTC):

<!-- cm:gallery --> <!-- /cm:gallery -->

What the numbers show [Verified, from the rows]

• —News coverage: "super intelligence" went from 0.007% of monitored online news on 20 Sep to 0.17% on 22 Sep and 0.22% on 23 Sep (about 30x); "superintelligence" rose from 0.02% (21 Sep) to 0.16% on 24 Sep. An earlier peak (0.14–0.15%, 9–12 Sep) follows coverage of AI-risk warnings by researchers, before the bill news.
• —Search interest (US): "super intelligence" rose from 4 to 65 on 22 Sep on the shared 0–100 scale, then 50 and 24 (24 Sep is a partial day). "AI ban" stayed at 11–15 through the week: the bill made news but not many people searched for "AI ban".
• —Hacker News: 32 unique stories in 30 days; the most discussed was the bill's announcement on 3 Sep (62 points).
• —Primary sources (checked 24 Sep 22:40 UTC): the Senate press release and the White House release are online and contain the quoted wording. No executive order or other presidential document in the 2026 Federal Register mentions superintelligence; the only 2026 Federal Register document that does is a 7 Jul FTC policy statement on AI accuracy. No bill number yet: the bill is not among the 250 most recently updated bills on congress.gov (titles often appear days after introduction).

Limits (read before using)

• —headlines is a sample, not every article: GDELT returned 2 of the 8 article-list windows we asked for (the others were rate-limited, HTTP 429), so 23 Sep and the "super intelligence" spelling before 21 Sep are well covered, other days less so. GDELT matches the article text, so some headlines don't contain the word itself (62 of 201 do). The daily coverage shares come from GDELT's full timelines and are complete (no data for 14–19 Sep in GDELT's timeline, left empty, never filled).
• —Google Trends values are a relative index, not search counts; "AI bill" hit 100 on 26 Aug, which sets the scale for all 5 terms.
• —Headlines are kept exactly as each outlet published them, including names and opinions in them; this dataset takes no position on the bill or the speech.

Use it

from datasets import load_dataset
daily = load_dataset("CyberMax-tools/superintelligence-ban-bill-coverage", "daily", split="train").to_pandas()
print(daily[["date", "news_share_pct_super_intelligence", "trends_super_intelligence", "hn_stories"]].tail(10))
news = load_dataset("CyberMax-tools/superintelligence-ban-bill-coverage", "headlines", split="train").to_pandas()
print(news.groupby("date").size().tail(10))

Columns

headlines: date, seen_utc (when GDELT first saw it / HN post time), title, outlet, language, source_country, url, source (gdelt or hacker_news), matched (search term), hn_points, hn_url.

daily: date; news_share_pct_superintelligence, news_share_pct_super_intelligence (% of all online news GDELT monitored that day); hn_stories; trends_superintelligence, trends_super_intelligence, trends_ai_ban, trends_ai_regulation, trends_ai_bill (Google Trends, US, 0–100).

sources: event_date, event, kind, url, http_status, confirmed_this_build, checked_utc.

Get fresh data

• —Search interest for your own topics, with a rising/falling verdict and alerts: Swellmeter: Google Trends scraper and API.
• —What people type about a topic: Keyfern keyword ideas; free snapshot for this story: superintelligence search questions.
• —Full article text for these links: Inkpluck article extractor (free demo: tech reads extracted).
• —Every federal AI executive order, rule and comment period, daily: Policywren US AI Policy Tracker.

Sources and licence

GDELT Project DOC 2.0 API (free, attribution: "The GDELT Project"); Hacker News via the Algolia HN Search API; Google Trends (public explore endpoints, US); Federal Register API (public domain); congress.gov API; the Senate and White House releases linked in sources. Headlines and links belong to their publishers and are listed as facts about coverage. Compilation CC BY 4.0 by CyberMax. Not affiliated with any source, office or party.

Refresh

python3 products/ai-policy-tracker/news/superintelligence-bill/build.py <out_dir> [gdelt_cache.json] (the optional cache holds GDELT responses fetched from another IP when this container's IP is throttled), then convert the three JSONL files to data/*.parquet and upload with this card. No Apify credit needed unless GDELT has to be fetched through Capturewren (~$0.02 per run).

<!-- cybermax-xlinks:start -->

More free from CyberMax

• —All free CyberMax demos & datasets
• —Policywren: US AI Policy Tracker (Federal Register, daily)
• —Swellmeter: Rising Google Searches in AI & Tech <!-- cybermax-xlinks:end -->

<!-- cybermax-notebook:start -->

Quickstart notebook

Load this dataset with pandas and see real outputs in the quickstart notebook. <!-- cybermax-notebook:end -->