It states that contributors "earn payouts for submissions that we approve" and publishes no per-hour rate, as checked on 2026-11-09. The company's terms, re-read on 2026-11-17, name the same entity as the contracting party.
Every figure here is a dated statement by Human Archive or by a named person speaking for it, handled by the same method as every buyer profile in this independent reference on AI training data licensing. The page records what is published, what is not and what the license keeps.
What Human Archive Collects and Who Licenses Its Datasets
Human Archive acquires new recordings made by its own contributors rather than existing datasets, and licenses the approved recordings to AI and robotics labs.
The data it asks contributors to record, per its privacy policy effective 2026-07-30:
- first-person video of everyday tasks
- voice exchanges
- depth and motion data
The route is recording to order under task rules. Human Archive describes itself as a "data and deployments infrastructure company for Physical AI".
No program for companies to license existing archives to it appears in the sources checked as of 2026-11-09, so the supply side is individual adults recording with the equipment and task rules Human Archive sets, and the demand side is the AI and robotics labs that license the result.
The distinction matters for a data holder: a company with a footage archive has no stated way in, while an individual contributor has a published, if rate-free, payout route.
How first-person footage is licensed and cleared, including wearers and bystanders, belongs to the page on egocentric first-person video for AI and robotics training; this page stays with Human Archive's own terms.
Human Archive Company Identity and Contracting Entity
The contracting party behind Human Archive is Human Archive, Inc., which its privacy policy describes as a Delaware corporation based at 466 Brannan Street, San Francisco.
The table lists the identity attributes on record as of 2026-11-09, with the privacy policy effective 2026-07-30 and the terms page (last updated 2026-07-30) re-read on 2026-11-17.
| Attribute | Value | Source |
|---|---|---|
| Legal name | Human Archive, Inc. | humanarchive.ai/privacy.html |
| Registry | Delaware corporation, file number not published | humanarchive.ai/privacy.html |
| Address | 466 Brannan Street, San Francisco | humanarchive.ai/privacy.html |
| Contracting party per terms | Human Archive, Inc., last updated 2026-07-30 (checked 2026-11-17) | humanarchive.ai/terms.html |
| Policy naming the entity | privacy policy effective 2026-07-30 | humanarchive.ai/privacy.html |
| Website | humanarchive.ai | humanarchive.ai |
| Country | United States | humanarchive.ai/privacy.html |
| Type | data vendor, capture lab | humanarchive.ai |
| Founding year | not stated on its primary pages | n/a |
| Founders and headcount | not recorded in the sources checked | n/a |
The terms page gives no postal address, only a contact email, so the address above rests on the privacy policy alone. The privacy policy and the terms carry the same last-updated date of 2026-07-30 and the same company name, which is why the contracting entity is stated without qualification.
No founding year, founders or headcount are public in the sources checked, and no funding amount appears on Human Archive's own pages checked.
Human Archive Contributor Program Terms and Payouts, Dated
Human Archive states that contributors "earn payouts for submissions that we approve", paid through Stripe and Circle, and publishes no per-hour rate; the table lists each term as of 2026-11-09.
Each value is Human Archive's own statement from its privacy policy unless the checked column says otherwise.
| Term | What Human Archive states | Checked | Source |
|---|---|---|---|
| Program | Human Archive contributor app | 2026-11-09 | humanarchive.ai/privacy.html |
| Data wanted | First-person video of everyday tasks, voice exchanges, depth and motion data | 2026-11-09 | humanarchive.ai/privacy.html |
| Payout | "earn payouts for submissions that we approve" | 2026-11-09 | humanarchive.ai/privacy.html |
| Payout per the terms | "Approved submissions earn the payout shown on the task; rejected submissions earn nothing." | 2026-11-17 | humanarchive.ai/terms.html |
| Per-hour rate | not published | 2026-11-09 | humanarchive.ai/privacy.html |
| Payout providers | Stripe and Circle | 2026-11-09 | humanarchive.ai/privacy.html |
| Reversal | Approval may be reversed | 2026-11-09 | humanarchive.ai/privacy.html |
| Process | Adult contributors record tasks under task rules; Human Archive reviews and approves | 2026-11-09 | humanarchive.ai/privacy.html |
| Eligibility | Adults; minors may not be the primary subject | 2026-11-09 | humanarchive.ai/privacy.html |
| Onward license | To AI and robotics labs under written agreements prohibiting re-identification | 2026-11-09 | humanarchive.ai/privacy.html |
| Retention and deletion | Kept for the license duration; not recalled on account deletion | 2026-11-09 | humanarchive.ai/privacy.html |
| Referral fee | none recorded | 2026-11-09 | humanarchive.ai/privacy.html |
| Program for companies with existing data | none on record | 2026-11-09 | humanarchive.ai |
The payout is a claim tied to approval: no rate, currency or timing is published in the privacy policy, and the terms say only that the payout is the one shown on each task.
How that compares with other buyers is set out in the program terms compared table, which places each claim next to its date.
Approval can be reversed after a submission is accepted, and payouts run through Stripe and Circle, so a contributor's earnings depend on the review outcome rather than on time spent recording.
The terms, read on 2026-11-17, say rejected submissions earn nothing, which is the other side of the same mechanism. Human Archive does not publish how long review takes or what share of submissions is approved.
On eligibility, the service is for adults, and minors may not be the primary subject of a recording. Neither a referral fee nor a program for companies with existing data is on record.
What Happens to Human Archive Recordings After Approval
Once Human Archive approves a recording, it licenses the recording to AI and robotics labs under written agreements that prohibit re-identification, and keeps it for the duration of that license.
The labs that receive the data are not named in the sources checked, so a contributor cannot tell from the public pages which developer ends up training on a given recording. The ban on re-identification is a term of the lab agreements, as Human Archive states it; the agreements themselves are not published.
An account can be deleted, but Human Archive states in its privacy policy that approved submissions are not recalled on account deletion.
Its terms, read on 2026-11-17, say the account and personal information are deleted while approved clips and voice recordings are de-identified and kept in its datasets. A contributor who changes their mind after approval therefore cannot expect the recording to leave the licensed datasets.
Contracts differ on recall; the page on deletion, return and machine unlearning clauses shows how those clauses are worded elsewhere and what each one commits the buyer to.
Human Archive Reputation: The Record as of 2026-11-09
No complaint, court case, regulator action or breach notice involving Human Archive appears in a primary or named-person source checked as of 2026-11-09.
The one figure from a named person is a claim: Raj Patel, speaking for Human Archive in a TechCrunch interview of 2026-05-26, stated more than 1,000 active headsets. The sources give no title for him, and the figure is not independently confirmed.
It describes the size of the recording fleet, not the number of contributors, the volume of data or any payout. No funding amount, investor or contributor country is stated here, because none appears on Human Archive's own pages checked.
Forum threads and social posts are not evidence and are not used. The dated log of data buyer incidents holds no Human Archive entry as of 2026-11-09.
Benefits, Risks, Downsides and Alternatives for Contributors and Data Holders Considering Human Archive
For a contributor, Human Archive offers published payout mechanics from a named US company, but no stated rate, reversible approval and no recall after deletion; for a company with existing footage it offers no route.
Benefits
- An identifiable US entity is named in its own policy and terms.
- Payout mechanics, payout providers and eligibility are published.
- Human Archive states that licensees are barred by contract from re-identification.
Risks
- Approval can be reversed.
- No rate is published, so the value of a recording is unknown before the task.
- Recordings are not recalled on account deletion.
Downsides
- No route exists for companies with existing footage.
- No referral fee is on record.
- The labs that receive the data are not named.
Alternatives (alphabetical, each as the party's own claim, checked 2026-11-09)
- Asimov: states it will "cover their salaries" for businesses whose workers wear recording kit.
- Claru: takes non-exclusive training rights; the partner keeps ownership.
- Kled: takes a transferable, perpetual and irrevocable license, exclusive for training third-party AI models.
- Sieve: states "You keep ownership"; no amount is published.
- Not recording or licensing at all.
The lists above are a reading aid rather than advice: whether a recording is worth submitting at an unpublished rate is for the contributor to judge against the terms in the table. The wider set of alternatives to selling data, including keeping it, is compared on its own page.
Human Archive Among Capture Labs and Dataset Vendors
Human Archive is one of the capture labs among the dataset vendors profiled here, the group that records new data to order instead of reselling existing collections.
It sits in the dataset vendors and data labs directory beside other vendors, and related pages cover the data types, the legal rules and the contributor side.
Human Archive differs from vendors that license existing archives in that its supply starts only after a contributor accepts a task, so the terms above are the main published description of how the data comes to exist.
- Who buys AI training data: every buyer and program
- Choosing a buyer for your data: labs, vendors, marketplaces and intermediaries
- AI training jobs and platforms compared: for contributors
- Pay for AI training work: dated pay by platform
- Is AI training work legit: how to check a platform
- Individuals selling their own data: options
- Robotics data for AI training: and who buys it
- Sensor data for AI training: depth and motion included
- Motion capture and 3D data: for AI training
- Selling your voice to AI: and likeness rights
- Biometric data in AI training: rules in force
- COPPA and children's data: in recordings
- Claru: also licenses existing footage
- Kled: a contributor app with an exclusive license
- Sieve: pays for captured real-world work
- Mecka: commissions robotics data collection
- Asimov: pays approved collectors
More About Human Archive
Frequently Asked Questions About Human Archive
Does Human Archive pay contributors?
Yes, as Human Archive states it: contributors earn payouts for submissions it approves, through Stripe and Circle, and approval may be reversed (privacy policy, checked 2026-11-09).
How much does Human Archive pay per hour?
Human Archive publishes no per-hour rate on its own pages checked as of 2026-11-09. Its terms say the payout is the one shown on the task (checked 2026-11-17).
Can I delete my Human Archive recordings?
An account can be deleted, but Human Archive states that approved submissions are kept for the license duration and not recalled on account deletion.
Which company runs Human Archive?
Human Archive, Inc., a Delaware corporation at 466 Brannan Street, San Francisco, per its privacy policy effective 2026-07-30. Its terms, last updated 2026-07-30, name the same company.
Can a company license its footage to Human Archive?
No program for companies to license existing footage appears in the sources checked as of 2026-11-09; Human Archive records new data through its contributors.