AI / Twitter & X
50 Best AI Twitter Accounts to Follow on X
A good AI feed needs more than launch announcements. Researchers explain what changed, engineers show what survives contact with real users, and reporters ask who benefits. These 50 AI Twitter accounts cover those different jobs.
The selection includes researchers, educators, builders, labs and policy voices. We have ranked the selected accounts by follower count so the order is reproducible. A smaller evaluation specialist may still be more useful to you than the CEO at the top of the table.
The 50 accounts
50 of 50 accounts · Ordered by follower count
- 01

Sam Altman
Industry leaders@sama
OpenAI announcements and the CEO's views on the direction of AI, products and society.
6,355,556followersView on X ↗ - 02

OpenAI
Labs & tools@OpenAI
Model releases, research and product announcements directly from the company. Read performance claims alongside independent tests.
5,419,222followersView on X ↗ - 03

Andrej Karpathy
Research & engineering@karpathy
Neural-network explanations, coding experiments and reflections on building with language models. Particularly useful for technically curious readers.
4,322,751followersView on X ↗ - 04

Demis Hassabis
Research & engineering@demishassabis
Research milestones and scientific applications of AI, including work connected to DeepMind and Isomorphic Labs.
1,937,330followersView on X ↗ - 05

Andrew Ng
Education@AndrewYNg
Applied machine learning, AI education and industry commentary from the Coursera and DeepLearning.AI co-founder.
1,919,108followersView on X ↗ - 06

Anthropic
Labs & tools@AnthropicAI
Claude releases, interpretability research and AI safety work from the company developing the models.
1,780,873followersView on X ↗ - 07

Google DeepMind
Labs & tools@GoogleDeepMind
Research papers, model announcements and scientific applications from the AI research organization.
1,563,897followersView on X ↗ - 08

Yann LeCun
Research & engineering@ylecun
Arguments about model architectures, world models and the limits of current AI. Often challenges prevailing industry assumptions.
1,322,128followersView on X ↗ - 09

Fei-Fei Li
Research & engineering@drfeifei
Computer vision, spatial intelligence and human-centered AI from the Stanford researcher and World Labs co-founder.
1,087,959followersView on X ↗ - 10

Greg Brockman
Industry leaders@gdb
OpenAI product and research developments from the company's president and co-founder.
1,068,234followersView on X ↗ - 11

Ilya Sutskever
Industry leaders@ilyasut
An occasional source of announcements and research perspectives from the Safe Superintelligence co-founder.
925,853followersView on X ↗ - 12
861,745followersView on X ↗ - 13

Dario Amodei
Industry leaders@DarioAmodei
Anthropic announcements and longer arguments about advanced AI capabilities, risks and deployment.
841,474followersView on X ↗ - 14

Hugging Face
Labs & tools@huggingface
Open models, datasets, demos and community projects across the machine-learning ecosystem.
812,950followersView on X ↗ - 15

Alexandr Wang
Industry leaders@alexandr_wang
AI industry strategy and developments connected to Meta's AI efforts and Wang's work founding Scale.
771,937followersView on X ↗ - 16

François Chollet
Research & engineering@fchollet
Reasoning, generalization and intelligence evaluation from the creator of Keras and ARC-AGI.
741,203followersView on X ↗ - 17

Geoffrey Hinton
Research & engineering@geoffreyhinton
Deep-learning perspectives and discussion of advanced AI risks from a foundational researcher in the field.
670,566followersView on X ↗ - 18

Rowan Cheung
News & discovery@rowancheung
A digest-style view of AI launches and applications. Useful for discovery before checking the underlying release.
600,599followersView on X ↗ - 19

Jim Fan
Research & engineering@DrJimFan
Robotics, simulation and embodied intelligence from a researcher working on NVIDIA's physical AI efforts.
596,694followersView on X ↗ - 20

Jeff Dean
Research & engineering@JeffDean
Machine-learning systems, computing infrastructure and research perspectives from a long-time builder of large-scale AI systems.
528,888followersView on X ↗ - 21

Sebastian Raschka
Education@rasbt
Technical explanations of language models and machine learning, with an emphasis on understanding implementations.
522,992followersView on X ↗ - 22

David Ha
Research & engineering@hardmaru
Research experiments, model design and developments from Sakana AI, where Ha is co-founder and CEO.
440,498followersView on X ↗ - 23

Ethan Mollick
Applied AI@emollick
Research-informed experiments on how AI changes work, education and organizational decisions.
400,703followersView on X ↗ - 24

DeepLearning.AI
Education@DeepLearningAI
Courses, explainers and learning resources for applied AI and machine learning.
352,801followersView on X ↗ - 25

NVIDIA AI
Labs & tools@NVIDIAAI
AI software, accelerated computing and applied research announcements from NVIDIA.
347,525followersView on X ↗ - 26

Jeremy Howard
Education@jeremyphoward
Practical deep learning, open models and accessible AI education from a fast.ai and Answer.AI co-founder.
330,251followersView on X ↗ - 27

Soumith Chintala
Research & engineering@soumithchintala
Open-source machine learning and robotics perspectives from a PyTorch co-founder.
327,701followersView on X ↗ - 28

Lilian Weng
Research & engineering@lilianweng
Research explanations and links to Lil'Log, covering topics such as learning methods and language-model behavior.
299,923followersView on X ↗ - 29

Gary Marcus
Policy & society@GaryMarcus
Critical analysis of generative AI claims, reliability and alternative approaches to machine intelligence.
251,614followersView on X ↗ - 30

Simon Willison
Applied AI@simonw
Hands-on model testing, developer tools and detailed notes on where AI systems work or fail.
235,672followersView on X ↗ - 31

Mistral AI
Labs & tools@MistralAI
Model releases, developer resources and product updates from Mistral.
221,582followersView on X ↗ - 32

Paul Couvert
Applied AI@itsPaulAi
AI and no-code workflows, tool demonstrations and ideas for putting models into everyday projects.
220,733followersView on X ↗ - 33

Shawn Wang
Applied AI@swyx
AI engineering practice, developer communities and interviews with people building models and applications.
200,774followersView on X ↗ - 34

Stanford NLP Group
Research & engineering@stanfordnlp
Papers, talks and research updates in natural-language processing from Stanford's research group.
195,130followersView on X ↗ - 35

Charly Wargnier
Applied AI@DataChaz
Open-source tools, agents and automation examples with a developer and data-science emphasis.
181,989followersView on X ↗ - 36

Timnit Gebru
Policy & society@timnitGebru
AI accountability, labor and research priorities from the founder of the Distributed AI Research Institute.
173,186followersView on X ↗ - 37

Chip Huyen
Research & engineering@chipro
AI engineering, evaluation and production systems from the author of books on building machine-learning applications.
153,112followersView on X ↗ - 38

Jack Clark
Policy & society@jackclarkSF
Research and policy analysis, with deeper reading through the Import AI newsletter.
150,945followersView on X ↗ - 39

Harrison Chase
Applied AI@hwchase17
Agent development, application infrastructure and LangChain ecosystem updates from its co-founder.
133,588followersView on X ↗ - 40

Yohei Nakajima
Applied AI@yoheinakajima
Public experiments with agents and small AI projects, including work associated with BabyAGI.
126,531followersView on X ↗ - 41

Cohere
Labs & tools@cohere
Enterprise AI, language-model releases and deployment resources, including open-model work.
121,399followersView on X ↗ - 42

Stanford HAI
Policy & society@StanfordHAI
Research, policy discussions and educational material about AI's effects on people and institutions.
109,463followersView on X ↗ - 43

Nathan Lambert
Research & engineering@natolambert
Open models, reinforcement learning and post-training research, with longer analysis through Interconnects.
104,134followersView on X ↗ - 44

Sebastien Bubeck
Research & engineering@SebastienBubeck
Research perspectives on language-model capabilities, reasoning and the mathematics behind machine learning.
93,279followersView on X ↗ - 45

Julien Chaumond
Labs & tools@julien_c
Product and community developments at Hugging Face, with attention to accessible machine learning.
89,315followersView on X ↗ - 46

Karen Hao
Policy & society@_KarenHao
Reporting on the companies, people and social costs behind AI development.
85,987followersView on X ↗ - 47

AI Now Institute
Policy & society@AINowInstitute
Policy research on AI accountability, industry power and the institutions shaping deployment.
58,934followersView on X ↗ - 48

Hamel Husain
Applied AI@HamelHusain
Practical guidance on evaluating AI applications, inspecting failures and improving systems beyond a polished demonstration.
58,627followersView on X ↗ - 49

Shreya Shankar
Research & engineering@sh_reya
Language models in data systems, evaluation and data processing, including work on DocETL.
58,009followersView on X ↗ - 50

Jay Alammar
Education@JayAlammar
Visual explanations of language models and their components, plus practical learning resources.
52,495followersView on X ↗
Build a feed that can disagree with itself
Start with a few primary sources such as OpenAI, Anthropic and Google DeepMind. Add an engineer who tests the releases, an educator who explains the underlying ideas and someone who scrutinizes the wider claims. Simon Willison, Sebastian Raschka and Karen Hao serve very different purposes in that mix.
For hands-on application work, look at Hamel Husain and Shreya Shankar for evaluation, Chip Huyen for systems thinking and Harrison Chase for agents. If your work is closer to robotics or computer vision, Jim Fan and Fei-Fei Li offer a more relevant starting point than a general AI news feed.
Read a model announcement past the benchmark chart
Open the paper, model card or repository linked in the post. Check what the evaluation measured, which version was tested and whether the result depends on a specific prompt, tool setup or extra computation. A compelling demonstration is evidence that something worked once, not proof that it is ready for your workflow.
Save examples of failures as well as successes. When a new release arrives, rerun a few tasks that matter to you: extracting information from your documents, following your output format or making changes in your codebase. The value of following experts is learning how to investigate the result, not inheriting their excitement.
How we selected and ranked these accounts
We selected accounts for identifiable work in AI research, engineering, education, industry reporting or policy. Public AI-account roundups helped surface candidates; profile records and available timeline samples were then checked separately. We kept a mixture of individual practitioners and research organizations instead of filling the list with general technology celebrities or virtual influencers.
Every selected account had at least 10,000 followers when checked.
We checked public profile records and, where available, sampled timeline results. Profile photos and follower totals are a snapshot collected on October 8, 2026 through the public FxTwitter API. Display names are edited for clarity. These are third-party readings of X profiles, not audited audience measurements. Numbers and handles can change after the check.
Entries are ordered by follower count, highest first. That ordering does not measure accuracy, audience authenticity, engagement or how often an account posts. The descriptions are editorial summaries; inclusion is not an endorsement of every post or a guarantee of future activity.
Sources and further reading
Each account links to its original X profile. We used the following sources for discovery and context, then checked profiles separately.