What 212 AI Creator Accounts Tell Us About Publishing in 2026
A snapshot of 212 AI creator accounts across nine platforms, with platform split, language split, topic split and the caveats that matter.
An AI creator, in this article, is a person or channel that regularly publishes about artificial intelligence on a public platform. This snapshot covers 212 platform accounts. Those accounts belong to 200 distinct names, because some creators run separate accounts on more than one platform.
The data comes from an internal deep-research report dated 13 October 2026. It used public platform pages, public directories such as Feedspot’s 2026 AI lists, and public statistics from SocialBlade. It did not use official platform APIs or logged-in scraping.
Where the accounts are
Accounts by platform, snapshot of 13 October 2026 (total 212)
| Platform | Accounts | Share |
|---|---|---|
| X (Twitter) | 52 | 24.5% |
| YouTube | 40 | 18.9% |
| Podcasts | 25 | 11.8% |
| 20 | 9.4% | |
| Medium | 20 | 9.4% |
| 15 | 7.1% | |
| Zhihu | 15 | 7.1% |
| Jianshu | 15 | 7.1% |
| note | 10 | 4.7% |
X and YouTube together make up about 43.4% of the records. The rest sit on podcasts, professional networks, long-form writing platforms and Chinese and Japanese content communities. If you only watch the two largest platforms, you miss most of the field.
Which languages
Of the 212 accounts, 172 publish mainly in English, 30 in Chinese and 10 in Japanese. The Chinese set leans on Zhihu and Jianshu. The Japanese set comes from note’s AI and machine-learning topic pages.
What they write about
Each account was given one main topic, so the counts below do not double up. The four largest groups, machine-learning engineering, AI research and science communication, generative AI, and applied case studies, hold 148 accounts, or 69.8% of the set.
Main topic per account (total 212)
| Main topic | Accounts | Share |
|---|---|---|
| Machine-learning engineering | 44 | 20.8% |
| AI research and science communication | 40 | 18.9% |
| Generative AI | 33 | 15.6% |
| Applied case studies | 31 | 14.6% |
| AI news and trends | 17 | 8.0% |
| Product and startups | 17 | 8.0% |
| AI ethics and governance | 14 | 6.6% |
| Data science | 9 | 4.2% |
| Robotics and vision | 4 | 1.9% |
| Education, security, chips (1 each) | 3 | 1.4% |
Largest accounts by public follower count
The ten largest accounts with a numeric follower or subscriber count are listed below. Each figure is the value shown by the named source on the date the report was compiled. Counts change daily, and the sources differ, so treat these as a snapshot, not a ranking you can quote as final.
Top 10 by public follower or subscriber count
| Rank | Creator | Platform | Count | Source named in report |
|---|---|---|---|---|
| 1 | 3Blue1Brown | YouTube | 8.69M | SocialBlade |
| 2 | Lex Fridman | X | 5.0M | Feedspot X list |
| 3 | Lex Fridman | YouTube | 5.0M | Feedspot YouTube list |
| 4 | Fireship | YouTube | 4.3M | SocialBlade |
| 5 | Sam Altman | X | 4.2M | Feedspot X list |
| 6 | Andrej Karpathy | X | 2.9M | Feedspot X list |
| 7 | Steve Nouri | 2M+ | GenAI.Works 2026 list | |
| 8 | Demis Hassabis | X | 1.6M | Feedspot X list |
| 9 | Andrew Ng | X | 1.6M | Feedspot X list |
| 10 | Pascal Bornet | 1.5M | Public LinkedIn creator list |
Medium follows the same rule. The Towards AI and Artificial Intelligence in Plain English entries are publications with many contributors, not individual writers. The report also lists TDS Archive, a Medium publication with about 828K followers according to its public page. That page shows its last post in 2025, so the report marks it as archived. A large follower count does not mean the feed is active.
What the numbers do not show
- Follower counts are not the same kind of number. A YouTube subscriber, an X follower, a podcast monthly-listener range and a Zhihu upvote measure different things.
- Some accounts show no public count. The report marks those as “not public” and does not estimate them.
- The dataset is not a global ranking of AI influence. It is a map of who publishes where, with the sources attached.
- LinkedIn was not crawled automatically. Its help pages prohibit third-party scraping, so LinkedIn entries rely on public profiles and public lists only.
How to use this
If you are looking for writers to follow on Medium, start with the topic and the last-post date, not the follower count. The next article in this blog, “How to read creator metrics without being misled”, explains the differences. The article after it gives a five-point check for a Medium AI writer.