Independent project. Not a U.S. government website.

Support our project (opens Buy Me a Coffee)
USASI

Jump to

Start hereFour ways into the catalogCompanies & LabsU.S. AI organizationsOpen Models & ToolsOpen models, software, datasets, and evaluation toolsLearnExplainers and learning pathsFor kidsAI and Super Intelligence, explained for readers 8 to 12Latest newsSourced news itemsPolicy trackerFederal AI laws, orders, and standards

Editorial log

Changelog

Genuine catalog and policy changes, newest first. Rebuilding the site is not a change and is never recorded here.

  1. PostHog visitor analytics and session replay

    The site now sends cookie-free page-visit, click, and page-leave analytics to PostHog, and may record anonymized session replays with form inputs masked. The Privacy page, FAQ, and security policy describe this alongside Cloudflare Web Analytics. USASI still sets no cookies and does not use localStorage.

    • PolicyAdd cookie-free PostHog page analytics and masked session replay; update the Privacy page

  2. Jobs pages moved to a separate site

    USASI no longer publishes a job board. Open roles now live on aipeoplewiki.com, and the old addresses forward there.

    • PolicyRemoved the jobs section; old addresses redirect to aipeoplewiki.com

  3. City jobs pages, starting with Memphis–Southaven

    Jobs now has a reusable city-hub page for metros that meet a listing threshold. The first hub is Memphis–Southaven (Memphis, TN and Southaven, MS), built from the same employer feeds as the rest of the board. Further cities are config entries; a hub below the threshold is a normal 404.

    • AddedCity jobs hub for Memphis, TN and Southaven, MS, linked from the jobs directory and the Tennessee and Mississippi state pages

  4. People pages moved to a separate site

    USASI no longer publishes people profiles. The People directory, People Behind Big SI, Voices of SI, and People Behind Local SI now live at aipeoplewiki.com, and the old addresses forward there.

    • PolicyRemoved the people, Big SI, Voices, and Local SI sections; their old addresses redirect to aipeoplewiki.com

  5. Jobs: how hiring works on top employer pages

    Company job pages for OpenAI, Anthropic, SpaceX, NVIDIA, and CoreWeave now include a short “How hiring works” note drawn only from each company’s official careers pages, stored as data so more employers can be added later.

    • AddedHow hiring works sections on the OpenAI, Anthropic, SpaceX, NVIDIA, and CoreWeave employer pages

  6. AI fellowships and postdocs on the jobs board

    A new jobs hub lists current fellowships, fellows programs, and postdoctoral roles from employers' own job feeds, grouped by employer. Fall is application season; the list is rebuilt from titles on each deploy, the same way internships are.

    • AddedJobs hub for fellowships and postdocs, linked from the jobs directory, and internships

  7. Rank badges for AI on GitHub and Hugging Face model pages

    Every AI on GitHub project page and every AI on Hugging Face model page now has a small rank badge maintainers can add to a README, linking back to that page. The image is a static SVG generated at build time from the same rank the page already shows. The badges page and the privacy note cover both the listed badge and the new rank badges.

    • AddedRank badges (/badges/github/<slug>.svg and /badges/huggingface/<slug>.svg) with Markdown and HTML on every GitHub project page and Hugging Face model page
    • UpdatedBadges for maintainers and privacy: rank badges as well as the listed badge

  8. AI on GitHub: the most-starred open-source AI projects, each with its own page

    A new section, AI on GitHub, ranks 9,507 open-source AI projects on GitHub by stars, from the public GitHub API. Each project has its own page with its description, stars, forks, language, license, topics, last push, related projects, a link to its USASI catalog record where there is one, and a generated hero image. The list can be searched and filtered by topic, language, license, and owner type. Link lists, courses and books, collected prompts and jailbreaks, and deepfake tools are left out by published rules; the methodology explains them. Stars measure popularity, not quality or safety.

    • AddedAI on GitHub (/github/), with a page per project; methodology updated

  9. AI on Hugging Face: the most-liked open models and datasets, each with its own page

    A new section, AI on Hugging Face, ranks 4,320 models and 836 datasets on the Hugging Face Hub by likes, from the public Hub API. Each has its own page with likes, downloads in the last 30 days, task, library, license, parameter count where the Hub reports one, gated access, base-model links for quantized versions and fine-tunes, links to its USASI catalog record and AI on GitHub page where there are ones, and a generated hero image. The lists can be searched and filtered by task, library, license, author type, gated access, and quantized versions. Items tagged not for all audiences, deepfake and undressing tools, "uncensored" or "abliterated" fine-tunes, and jailbreak collections are left out by published rules; the methodology explains them. Likes measure popularity, not quality or safety.

    • AddedAI on Hugging Face (/huggingface/ and /huggingface/datasets/), with a page per model and dataset; methodology updated

  10. AI jobs on Hugging Face: a searchable snapshot and explorer

    Open listings from the USASI jobs board are published as the Hugging Face dataset usasihq/usasi-ai-jobs, with a Gradio explorer that filters them and links every row back to its page on this site. The daily jobs refresh uploads a fresh CSV when a Hugging Face token is configured.

    • AddedHugging Face dataset usasihq/usasi-ai-jobs and Space explorer; daily refresh can push the snapshot

  11. Hugging Face jobs explorer is a free static Space

    The USASI AI Jobs explorer on Hugging Face now runs as a static Space (plain HTML and JavaScript in the browser). Gradio on free cpu-basic needs a PRO subscription; static Spaces do not. Filters and links to this site’s job pages are unchanged.

    • UpdatedHugging Face Space usasihq/usasi-ai-jobs rebuilt as a free static explorer

  12. Jobs: weekly AI Hiring Index

    A public AI Hiring Index compares open listings by employer and U.S. state week over week, next to the WARN Act layoff pages. Each Jobs refresh stores a small dated count snapshot so the first week stands on its own and the trend begins the week after.

    • AddedAI Hiring Index: weekly open-job counts by employer and state, week-over-week change, and WARN Act pairing

  13. T3 Code and Ping Labs in the catalog

    Added T3 Code, the open-source app from Ping Labs (T3 Tools, Inc.) for running coding agents, and a Ping Labs organization record. TypeSafe AI's record now lists Jev 1.14 and its agent skill.

  14. Weekly catalog upkeep: Muse Spark Open, LFM2.5-24B-A2B, Nemotron 3.5 Super VL, Dory1.0-4B, and Granite Switch 4.1 8B

    Added five open-weight releases from the past week by catalog organizations, each under an existing family record: Meta's Muse Spark Open (Muse Model License, gated download), Liquid AI's LFM2.5-24B-A2B (LFM Open License v1.0), NVIDIA's Nemotron 3.5 Super VL 120B-A12B and Nemotron Labs Dory1.0-4B (OpenMDW License 1.1), and IBM's Granite Switch 4.1 8B (Apache 2.0). A batch of 40 cited sources was re-checked; all were first snapshots, so no changes could be compared, and no record needed a correction.

  15. Remote, hybrid, and on-site labels from employers; all of NVIDIA's listings; four universities

    When a job board leaves the workplace blank, Jobs now fills it only from what the employer states: every location given as remote, Workday's own remote-type field, or an explicit sentence about the role in the description. Listings that state nothing stay unlabeled. Workday lists that stop at 2,000 are now read in parts, bringing in all of NVIDIA's openings, and Carnegie Mellon University, Cornell University, The University of Texas at Austin, and the University of Washington are added.

  16. Federal AI jobs from USAJOBS.gov

    The jobs board now includes U.S. federal civilian openings related to artificial intelligence, machine learning, and data science, collected from the official USAJOBS Search API. They appear under a federal parent listing, named by hiring agency, with apply links on USAJOBS.gov.

  17. Jobs by role, starting with Forward Deployed Engineer

    The jobs board now has pages by role, beginning with Forward Deployed Engineer. Counts come from the same employer feeds as the rest of the board, matched on job title only.

    • AddedJobs by role, starting with Forward Deployed Engineer; later roles are new config entries

  18. Jobs: search first, filters by category, level, and pay, 15 role pages, and five more employers

    The Jobs page now opens with a search for what and where (a city, a state, or "Remote"), one-tap filters for remote U.S. work, entry-level roles, internships, and disclosed pay, and a filter panel with job category and experience level (read from the job title), pay range, and date posted, each with live counts. Results show pay, how long ago the job was posted, and an Apply link, and can be sorted by newest or highest pay. Role pages now cover software, machine learning, research, data, product, program management, solutions, sales, security, infrastructure, hardware, data center, design, and recruiting. Every job page lists similar jobs. Listings from Block, Palantir, Replit, METR, and Epoch AI are now collected, including Lever job boards.

    • UpdatedJobs directory: what/where search, quick filters, category, level, pay, and date filters with counts, pay sort, and result rows with pay and posting age
    • AddedFourteen more role pages on the jobs board
    • AddedSimilar jobs on every job page
    • UpdatedBlock
    • UpdatedPalantir
    • UpdatedReplit
    • UpdatedMETR (Model Evaluation & Threat Research)
    • UpdatedEpoch AI

  19. Jobs from ServiceNow, Lawrence Livermore National Laboratory, Hugging Face, Nomic AI, and FAR.AI

    Jobs now reads Workable and SmartRecruiters job boards as well, adding listings from ServiceNow, Lawrence Livermore National Laboratory, and Hugging Face, plus Nomic AI and FAR.AI from their Ashby boards. Each listing has its own page with the employer's description.

  20. Jobs from SpaceX and Anduril

    Open positions at SpaceX and Anduril Industries are now collected from their Greenhouse job boards, adding about 5,300 listings to Jobs, each with its own page. To make room within the hosting limit on files per deployment, internal links are now ordinary page links and the site no longer publishes its page-transition data files.

    • UpdatedSpaceX
    • UpdatedAnduril Industries
    • UpdatedInternal links load pages directly; page-transition data files are no longer published

  21. Jobs from NVIDIA, Micron, Salesforce, Intel, and six more employers on Workday

    Jobs now reads employers' Workday career sites, adding open positions at Adobe, Boston Dynamics, Broadcom, Intel, Marvell, Micron, NVIDIA, Red Hat, Salesforce, and SiFive. Each listing has its own page; descriptions are added over the first few refreshes.

  22. A Jobs page for every employer and for every current listing

    Each employer with an automated job feed now has a shareable page, and every current listing has its own page with the employer's description as published on its job board, the key facts, and a link to apply on the employer's site. Job search understands common abbreviations such as ML, SWE, and PM, ranks title matches first, and gained popular searches and a city filter. Listings are now refreshed twice a day.

    • AddedEmployer jobs pages, linked from the jobs directory and each organization page
    • AddedA page per current listing with the employer's description, apply link, and job posting data for search engines
    • UpdatedJobs search: abbreviations, title-first ranking, popular searches, and a city filter

  23. Twelve more documented local setups for the finder and the memory pages

    The finder now records twelve more local configurations, each with a memory figure from the model or runtime publisher's documentation for one exact variant: Gemma 4 12B, 26B A4B, and 31B in Google's QAT Q4_0 GGUF builds, Muse Glimmer 30B, gpt-oss-20b, Olmo 3 7B Instruct, Rnj-1 Instruct, Llama 3.1 8B, StarCoder2, and Whisper large-v3, run with llama.cpp, LM Studio, Transformers, or the openai-whisper package. Each entry quotes what its figure measures, such as a minimum, a loaded-model footprint, or a file size. The pages for 8, 12, 16, and 24 GB of VRAM and for 8, 16, 24, and 32 GB of Apple silicon memory now list different sets. None of these setups has been tested by USASI.

    • AddedTwelve finder configurations with publisher-documented memory figures, spanning chat, documents, code, and transcription

  24. Local AI by memory: what fits 8 to 24 GB of VRAM or an Apple silicon Mac

    New fixed pages list the documented local configurations from the finder that fit 8, 12, 16, or 24 GB of graphics memory, or 8 to 64 GB of Apple silicon unified memory, by each publisher's own memory figure. Each entry names the exact variant, quotes the publisher's statement with its source, and links to the catalog record and to the finder pre-filled for that hardware. The pages use the finder's own rule and say what they do not cover: speed, context length, and untested setups.

    • AddedPages at /find/memory/ for 8, 12, 16, and 24 GB of VRAM and 8, 16, 24, 32, and 64 GB Apple silicon Macs, linked from the finder and the local AI hub

  25. A larger, filterable news section

    Latest news gained items for July through October 2026 from official sources, a category, organization, and month filter, and an event-date order.

    • AddedNews items for events from July 1 to October 12, 2026; news filters and month headings on /news/

  26. A page for every document in the policy tracker, starting with Executive Order 14434

    Each law, executive order, memorandum, framework, and report in the U.S. AI policy tracker now has its own page with the recorded summary, key provisions, dates, issuer, status, related documents, and the same cited sources. The tracker keeps its overview and its old links to each entry, and every entry now links to its page.

    • AddedPer-document pages at /policy/<slug>/; site search now opens these pages for policy results

  27. The State of American AI, a monthly report, and Jobs pages by state, city, and topic

    A new report, The State of American AI, is computed from the catalog and the Jobs data and published once a month as a permanent edition with downloadable tables, starting with October 2026 at /report/2026-10/. Jobs gained pages for each U.S. state and for cities with enough listings, plus topic pages for remote, entry-level, no-programming, and data-center work, each built from the employers' own job boards.

    • AddedThe State of American AI at /report/, with the October 2026 edition at /report/2026-10/
    • AddedJobs pages by U.S. state, city, and topic (later moved off this site)

  28. Preview cards for every page and RSS feeds for every company, model, and topic

    Links to records, news items, explainers, hubs, license guides, and weekly issues now show their own preview card when shared, with the page title and key catalog facts. New RSS feeds cover each organization, each open model or tool, each topic hub, the weekly email, and the changelog, listed on a new Follow USASI by RSS page.

    • AddedPer-page social preview cards (1200×630) with descriptive alt text for 522 pages
    • AddedRSS feeds for every organization, open model or tool, and hub, plus the weekly email and the changelog; feeds page at /feeds/
    • UpdatedChangelog entries now have stable anchors, so feed items link to the exact entry

  29. In-depth pages for xAI, Tesla, Meta, Anthropic, and OpenAI

    The organization pages for xAI, Tesla, Meta, Anthropic, and OpenAI gain an in-depth guide with an overview, themed sections, a models and products lineup, a dated timeline, and leadership, each line linked to its official source.

  30. The catalog as a dataset on Hugging Face

    A dated snapshot of the published organizations, open models and tools, and news items is now available as CSV tables on Hugging Face (usasihq/usasi-catalog), with a dataset card that explains how the catalog is made and its limits. The Data and reuse page links to it.

    • AddedHugging Face dataset usasihq/usasi-catalog: artifacts, organizations, and news as CSV (CC BY 4.0)

  31. AI and Super Intelligence, explained for kids

    A new section for readers about 8 to 12, meant to be read with a grown-up: seven short lessons on what AI is, how it learns, what "super intelligence" means (both the new U.S. government name and the older idea), what AI can and cannot do, using it safely, how people keep it safe, and AI in future jobs. Each lesson has an activity, a quick quiz that saves nothing, and a sources note for grown-ups. A word list and a note for parents and teachers go with them.

    • AddedFor kids section: /kids/, seven lessons, a word list, and a page for parents and teachers

  32. Search from any page, a states map, and printable kids lessons

    A search button in the header (or the / key, or Ctrl or ⌘ K) now searches every published record and page from anywhere on the site: organizations, open models and tools, explainers, learning paths, kids lessons, news, policy, licenses, the glossary, and the site's own pages, with a shareable results page at /search/. Places now opens with a tile map of headquarters states. Kids lessons print as worksheets with an answer key, and a classroom packet puts all seven on one printable page. Every page now prints in dark ink on white without the site header and footer.

    • AddedSite-wide search dialog, /search/ results page, OpenSearch description, and search suggestions on the not-found page
    • AddedTile map of headquarters states on /places/
    • AddedPrint styles for every page; printable kids lessons and a classroom packet at /kids/packet/

  33. Share and cite tools, a badge for maintainers, and an archive of the weekly email

    Catalog records, news items, explainers, and hubs now have plain share links, APA and BibTeX citations, and a link to the weekly email. Maintainers of listed projects can add a small "Listed in USASI" badge that links to their entry. Sent issues of the weekly email are now archived on the site exactly as they were emailed, starting with issue 1. The privacy page explains share links, copy buttons, and the badge.

    • AddedShare or cite this page: links to share on X, Bluesky, LinkedIn, Reddit, Hacker News, or by email; APA and BibTeX citations
    • AddedBadges for maintainers (/badges/): a "Listed in USASI" badge with Markdown and HTML on every organization and open model or tool page
    • AddedWeekly email archive: issue 1 at /newsletter/issue-1/, listed on /newsletter/
    • UpdatedWeekly email signup added to record pages, explainers, hubs, the finder, and the page-not-found page
    • UpdatedPrivacy: share links, copy buttons, and badge image requests

  34. The USASI weekly: a page for the weekly email

    New page explaining the weekly email: what each issue contains, how issues are assembled from material already published on the site, when they arrive, how signup and unsubscribing work, privacy, and sponsor labels. The privacy page now says that the email provider measures opens and link clicks.

    • AddedWeekly email page (/newsletter/), linked from the footer, the mobile menu, and every signup box
    • UpdatedPrivacy: the email provider measures whether emails are opened and which links are clicked

  35. Weekly catalog upkeep: AlphaProtein Novo and Liquid AI d1

    Added two open releases from the past week by catalog organizations: Google DeepMind's AlphaProtein Novo enzyme-design pipeline (code Apache-2.0; generator weights under non-commercial terms) and Liquid AI's d1-3B decision model (LFM Open License v1.0), each with a family record. A batch of 40 cited sources was re-checked; one license page had moved, and its URL was updated in two records.

  36. Learn becomes a learning center: 12 new explainers, learning paths, 35 glossary terms, three hubs

    Learn now has 22 sourced explainers grouped into six topics, five learning paths, key takeaways and an on-this-page outline on every explainer, and a glossary of 69 terms with an A to Z index. Three new topic hubs cover safety and security, developer tools, and speech, vision, and multimodal AI. Each new page was written from primary sources read on October 8, 2026, and then checked by a separate fact-checker against those sources.

    • AddedExplainers: how language models work, tokens and context windows, multimodal models, a short history of AI in the United States, pretraining and post-training, quantization
    • AddedExplainers: how models use tools, retrieval-augmented generation, labels and content credentials, safety testing and frameworks, careers in AI, what happens to what you type
    • AddedLearning paths: New to AI, Choosing and running a model, Building with AI, Checking claims about AI, Understanding the industry
    • AddedHubs: safety and security; developer tools; speech, vision, and multimodal
    • Added35 glossary terms, from agent and attention to vector database and watermark
    • UpdatedLearn page, explainer pages, glossary, hubs index, and homepage redesigned; Learn added to the main menu

  37. Eight new catalog records and an EmbeddingGemma note

    Added Google DeepMind's EmbeddingGemma 2 and SynthID Text, Meta's Llama Guard 4 and Llama 3.1 8B, Ai2's OLMo-core, OpenAI's tiktoken, AMD's ROCm, and Sentence Transformers. Several of these were already named in the explainers and reference hubs without a record of their own. The EmbeddingGemma 300M record now notes that Google lists EmbeddingGemma 2 as a newer model.

  38. Tools for deciding: a finder, a license guide, a policy tracker, an FAQ, and quizzes

    New "Find AI for my needs" finder matches a task, computer, comfort level, and willingness to pay against 17 documented configurations and explains every match, including what is not verified and that USASI has not tested them. Also new: plain-language guides to 20 licenses, a tracker of 30 U.S. federal AI policy documents with their current status, a 28-question FAQ, a model size calculator, "check your understanding" questions on every explainer, job tracks for people who are not AI engineers, and filtered CSV and JSON downloads in both directories.

    • AddedFind AI for my needs (/find/): 17 configurations across writing code, working with documents, transcribing audio, and chat
    • AddedLicense guide (/licenses/): 20 licenses, each linked from the records that use it
    • AddedU.S. AI policy tracker (/policy/): 30 federal laws, executive orders, OMB memoranda, and NIST publications
    • AddedFAQ (/faq/), model size calculator, and "check your understanding" questions on all 22 explainers
    • UpdatedJobs: four tracks and a label for postings whose titles present them as general expressions of interest
    • UpdatedHomepage task entry points; review levels defined in the methodology; each explainer says how it was made
    • UpdatedWeekly email now links to the USASI weekly on beehiiv

  39. Hermes Agent, Hermes, and Nous Research added to three hubs

    The Agents hub now features Nous Research and its Hermes Agent platform, with a new passage on agents that keep memory, skills, and scheduled tasks between conversations. The Local AI hub now features Hermes Agent as an example of an agent that can run on a self-hosted model, and explains that the model and the commands an agent runs can be set up separately. The Foundation Models hub now features Nous Research and the Hermes model family, with a passage on model families built on another developer's base model.

    • UpdatedAgents hub: added Nous Research and Hermes Agent as featured records and a sourced passage on Hermes Agent
    • UpdatedLocal AI hub: added Hermes Agent as a featured record and a sourced passage on running agents with local models
    • UpdatedFoundation Models hub: added Nous Research and the Hermes family as featured records and a sourced passage on models built on another developer's base model

  40. Weekly catalog upkeep: three new open releases

    Added three open-source AI projects released in the past week by catalog organizations: Meta's SWE-sweep coding-agent benchmark, Google DeepMind's SynthID Bio protein watermarking code, and OpenAI's MCP Extensions SDKs. A batch of 40 cited sources was re-checked; none showed a material change, so no existing record was edited.

  41. Five more hubs and new organizations

    Added five subject hubs (foundation models, robotics, evaluation, enterprise AI, and data), 10 organizations and 8 open artifacts, and structured data for explainers and the glossary.

  42. Corrections from source re-checks: Gemma 4, Olmo 3, Nemotron 3, EmbeddingGemma, Granite 4.2

    Re-checked flagged records against their primary sources. Updated the Gemma 4 license link to its current page, added training-data information assessments to two Olmo 3 releases and training-data access assessments to three Nemotron 3 releases, and recorded conflicting access and context-length statements for EmbeddingGemma and Granite 4.2 as the sources state them.

  43. Source library, timeline, and search across the whole reference

    Added a searchable source library of every source cited on the site, with the records that cite each one; a timeline of open releases and news events built from documented dates; and site search that now also finds hubs, explainers, glossary terms, and states.

    • PolicyNew pages: /sources/ and /timeline/; search now covers reference pages

  44. Eligibility decisions: llama.cpp, MLC LLM, and Lemonade; SpaceXAI jobs

    Published three local-AI projects whose U.S. maintaining organization is now documented: llama.cpp (maintained by the ggml team at Hugging Face), MLC LLM (a Carnegie Mellon Catalyst research project), and Lemonade (backed and presented as its own by AMD). Added the SpaceXAI job board to the xAI record.

  45. Correction to the GPT-6.1 Sol news item

    The item said one changelog entry added computer use and an "ultrafast" service tier; OpenAI lists these as separate entries on the same date, and the GPT-6 Astra feature is called Ultrafast mode.

    • CorrectedNews item "OpenAI adds GPT-6.1 Sol to its API": wording corrected against the OpenAI API changelog

  46. Orientation pages, a glossary, and the first explainer

    Added a persistent independence line to every page, a Start here guide, a 22-term glossary, an explainer on what open weight and open source actually mean, and a data-and-reuse page for the existing exports. Jobs now links to verified official careers pages for employers without a supported feed and collects 14 more employer feeds.

    • PolicyNew pages: /start/, /glossary/, /learn/open-weight-vs-open-source/, /reuse/

  47. Reference build-out: hubs, ten explainers, places, and record context

    Added five subject hubs, a Learn section with ten explainers, organizations by headquarters state, a "What this catalog does not know" section and related links on every record, comparison export (CSV and JSON with caveats), a larger glossary, and a pilot of national laboratories and universities. The About and Contribute pages now state that USASI names no individual editor and promises no response time for corrections.

    • PolicyNew sections: /learn/, /hubs/, /places/; comparison export on /matrix/
    • PolicyEditorial accountability: no named editor; corrections reviewed as time allows with no promised response time

  48. Round 3: agent tools, local AI, infrastructure, science, and deeper profiles

    Added 29 organization records and 55 open-artifact records from official, public sources: open agent protocols and SDKs (Model Context Protocol, A2A, AGENTS.md, goose, Codex CLI, Gemini CLI, and others), tools for running models locally, AI infrastructure companies, developer platforms, universities, NIST, and science-model makers, including TypeSafe AI. Added profiles for ten large AI companies, eleven news items, and job feeds for 28 more employers.

  49. Compact v0.2 and an optional homepage sponsor panel

    Revised the sponsorship policy, Privacy page, and USASI Compact to permit one clearly labeled homepage advertisement below the directory previews and before news, in addition to weekly-email sponsorships. The panel stays hidden without a configured sponsor. Free access, editorial independence, and the ban on paid listings, rankings, badges, and reader-data sharing remain. This revision does not change eligibility criteria or openness rubric v0.1.

    • PolicyUSASI Compact v0.2: one independent homepage advertisement, with disclosure and no tracking
    • PolicyOptional sponsor panel hidden by default; tips remain separate

  50. Separate component rights, data disclosure, and access

    Rubric v0.2 separates training-data information completeness from access and requires an explicit sourced system-rights review for the highest tier. Legacy data assessments remain visible but do not establish the new completeness checks. Added component-specific license filters, clearer hosted-product labels, fact-level review fields, deeper representative company profiles, email corrections, and a manual source-change proposal queue. Existing review dates are not refreshed by migration.

    • PolicyUSASI rubric v0.2; Open system (reviewed) replaces the visible Fully open label; no OSI certification claimed
    • PolicyComponent matching, profile evidence, review queue and account-free email corrections

  51. Cloudflare Web Analytics privacy disclosure

    Updated the Privacy page to explain optional cookie-free Cloudflare Web Analytics before activation. The security policy permits Cloudflare's measurement script while retaining same-origin collection. No advertising pixels, paid tier, or reader-data sharing with sponsors is introduced.

    • PolicyDisclose Cloudflare Web Analytics and its collection before activation

  52. Expansion: tech giants, open-source makers, universities, and the Local corner

    Added 57 organization records and 122 open-artifact records, each fact-checked against its sources: U.S. technology companies, AI application companies, open-model labs, robotics companies, open-source tool makers, foundations, universities, and big-tech open releases, plus benchmarks and datasets. 3 historical records were added as archived entries. Added official contact and profile links, and new schema values for university labs, standards bodies, public institutions, and science.

  53. Initial catalog

    First published set of reviewed records: 42 organization records and 56 open-artifact records, each fact-checked against its cited sources before launch. 3 historical records were added as archived entries. The eligibility policy, openness rubric, and USASI Compact were published at version 0.1.

  54. Latest news, the weekly email, and a sponsorship policy

    Added Latest news: short, dated, sourced updates about organizations and artifacts in the catalog, linked from the homepage, each entry page, and an RSS feed. Added a signup block for the upcoming USASI weekly email (hosted by the newsletter provider; the site collects no email addresses) and published a sponsorship policy: sponsorships appear only in the weekly email, are labeled, and never affect the catalog. The homepage artwork is now smaller on desktop so the introduction, search, and both directories are visible without scrolling.

    • PolicyLatest news section and RSS feed
    • PolicySponsorship policy (newsletter only, labeled, no influence on the catalog)

Support Us

Help keep USASI useful.

Leave a tip on Buy Me a Coffee (external site)

Optional. No USASI account required. Payment takes place on the linked provider’s website (Buy Me a Coffee).

About supporting this project