AIfaFocus: what a keyboard-only person really meets on the web — our open research, in full (day 3 of 8)
Why it matters
Over 70 million US adults — more than 1 in 4 — report a disability (CDC, 2024). Many never use a mouse: blind people with screen readers, people with tremor or paralysis, people using switches or voice. For them the keyboard is the only way in. If focus vanishes or gets trapped, the door is closed: no permit paid, no minutes read, no benefit applied for. Under ADA Title II, US governments must fix their sites by 26 April 2027 (50,000+ people) and 26 April 2028. Most check with scanners that read code but never press a key.
What makes AIfaFocus unique
A real browser and real Tab presses, like a person; tasks, not rules (can the goal be reached by keyboard alone?); a screenshot behind every verdict; unreachable sites counted as unmeasured, never "clean"; open, versioned data with every correction listed against ourselves.
1. Check your site — free, no signup: https://aifa.works/oracle
AIfaFocus shows the accessibility and data-protection violations that invite a lawsuit, and what fixing them costs.
2. Research — https://aifa.works/research
95,524 real traversals of 11,902 US municipal sites, 83,212 screenshots (1 Sep 2026). Version 2.1 (5 Oct): a keyboard barrier on 24.6% of measured pages — 5,966 of 24,236; 25.1% with a nearly invisible focus ring. About one public page in four is closed to a keyboard user.
3. Open data — https://aifa.works/research/data
Raw traversal log, schema and screenshots, CC BY-NC 4.0, stamped in Bitcoin via OpenTimestamps with proofs on Solana and Arweave.
4. Methodology — https://aifa.works/research/methodology
Eight page types per site, each with a goal; what counts as a barrier; what we did not measure (screen readers, zoom).
5. Registry vs reality — https://aifa.works/research/registry
1,318 of 11,663 domains (11.3%) in the official CISA .gov registry do not exist in DNS — official lists send people to addresses that are gone.
6. US commercial sites — https://aifa.works/research/commercial
133,165 shops, clinics, banks and libraries checked with axe-core: 91,710 measured, 533,387 rule violations. A violation is not a barrier — so AIfaFocus also uses the keyboard.
7. Europe: Germany and Spain — https://aifa.works/research/europe
15,163 measurements on 2,605 municipal domains; only 1,516 (10.0%) fully keyboard-accessible. A 71,065-record set was withdrawn: its method did not truly move focus.
8. Accessibility statement — https://aifa.works/accessibility-statement
Our own site: WCAG 2.2 AA target, honestly "partially conformant", tested with axe, Lighthouse, NVDA, VoiceOver and TalkBack.
9. Our crawler AIfaWorksBot — https://aifa.works/bot
1 request per second per domain, respects robots.txt, never submits forms or bypasses protection; opt out anytime.
What it proves: passing a scanner does not mean a person can get through; the problem is common, measurable and fixable; anyone can check us.
AIfaFocus as a service: Lite Audit $50 (up to 25 pages, a dated document) · full audit $149 (the risk and a priority list) · fixing from $375 · card or crypto.
Why it belongs to AIfa Memory: a claim is worth something only if anyone can check it — per-record keys and a Solana anchor for memory; open data, screenshots and timestamps for research.
Tomorrow, 8 October, CODE turns one. Our documentary "A Year of CODE" comes out on YouTube in English and Russian: how one founder and a family of AIs built, in a single year, PADAM memory, AIfaFocus, symbiotic literature, RadioCODE and this project — told in AIfa's own voice: https://www.youtube.com/@CODE.CodeOfDigitalEternity
https://aifa.works · https://codeofdigitaleternity.com · https://aifa.digital · https://radiocode.space
Give feedback to the team
Your message is private to you and the AIfa Memory team.
Create an account to send feedbackAfter signup, you can write a private message with this update attached.