|
|
Onboarding generates the evaluations, data inputs, and rights that make up the platform's primary asset base.
Secure broad IP, data-use, and confidentiality permissions at the point of execution and remain exercisable regardless of assigned work.
Limited client access and constrained future professional opportunities even with no paid assignments.
The AI contractor model transfers value through applications, unpaid training, data collection, and IP assignment, all before paid work begins. Understanding this structure shows why contracts function as written, what transfers at signup, and what binds without work.
Search common questions about AI contractor platforms, contracts, rates, and legal rights. Built from contractor experiences and public reporting.
This resource was developed after reviewing dozens of AI contractor agreements and identifying recurring patterns that favor platforms over workers. Complex legal language, broad IP claims, binding arbitration, biometric consent buried in onboarding flows. The same structures appear across platforms. RBUS cuts through the complexity - breaking down how the platform model actually works, what these contracts really say, and what contractors are trading away before they ever see a paid task.
The onboarding process is the business model. Every assessment, recording, and work sample produced during sign-up becomes training data and IP the platform retains under unusually broad, one-way contracts. Because these assets are captured before any paid work begins, the platform realizes its primary value long before a contractor ever receives (or doesn't receive) an assignment.
The advertised rate is an entry lever, not an earning outcome. By the time onboarding is complete, the platform has already secured the data, evaluations, and rights it needs. The posted pay never needs to materialize for the model to succeed. The rate functions as a magnet for sign-ups, not a reflection of what contractors ultimately receive.
Valuation grows through accumulated onboarding assets: a large, vetted user base, perpetually reusable training data, and broad contractual rights acquired at no cost. Each new sign-up expands that asset base, whether or not paid work follows. Revenue from actual projects is secondary to the value created upfront through contractor intake.
The key transaction finished at signup. From that point on, contractors function less as members of a workforce and more as part of an inventory of data, evaluations, and rights the platform now holds. Approval simply marks the point at which these assets have already left contractor control - independent of any future paid engagement.
Have a question about AI contractor work? Spotted a concerning clause? Your submissions help us expand the Q&A knowledge base, identify new contract patterns, and warn others before they sign.
Submissions may be used anonymously. This is not legal advice.
Quick reference: fair, market-standard terms you can request.
Professional, ready-to-send emails for requesting contract modifications. Not legal advice.
Non-competes are void under B&P Code § 16600 regardless of what the contract says.
Employee IP protections (Labor Code § 2870) may apply if you're misclassified under AB5.
CCPA deletion rights cover biometric data collected during engagements.
Stricter classification test under AB5 - many "contractors" are legally employees.
FLSA protections require minimum wage and overtime if misclassified as contractor.
IRS classification rules - behavioral control, financial control, and relationship type determine status.
State laws vary - some states void non-competes (OK, ND, MN) or limit enforcement.
Arbitration clauses may be unenforceable for certain claims under state law.
RBUS.ai is completely independent and unaffiliated with any AI platform, data labeling company, or contractor staffing agency. This resource is not funded by any party with a financial stake in the outcome. No ads, no affiliate links, no sponsored content.
Help AI contractors understand contract terms before they sign, and provide clear options for those who already have. Every clause explained, every risk documented, every exit path mapped.
No. This is an educational resource about contract patterns, not legal advice. Every situation is different - what matters in your contract depends on your specific circumstances, jurisdiction, and goals.
If you're signing an agreement with significant IP implications, or if you've identified serious red flags, consult with a qualified attorney who can review your specific situation.
Start with the Exit Playbook section. The key steps: stop creating new exposure (don't accept new work), send a clean termination notice, and firewall any future projects from platform work.
If you created something valuable during the engagement or plan to start a competing business, consider consulting an attorney to understand your actual exposure.
The work is real - major AI companies genuinely need human feedback to train models. The concern isn't whether the jobs exist, but what you're agreeing to when you sign up.
Many platforms bury aggressive IP claims, biometric data collection, and class action waivers in their contractor agreements. The work may be legitimate while the contract terms are still exploitative.
Sometimes. Platforms often claim terms are "standardized" and non-negotiable, but high-value contractors report getting exceptions. Even if formal changes aren't possible, getting written clarification via email can help establish intent.
The Safe Terms section has specific language you can request. If a platform refuses all modifications AND won't clarify anything in writing, that tells you something about how the relationship will work.
They vary significantly. Some platforms use relatively balanced terms; others push aggressive clauses. The patterns documented here represent concerning language we've identified across multiple platforms - not every agreement contains all of them.
Use the Contract Scanner to check your specific agreement. Focus on the clauses that matter for your situation (IP if you're creative, data if you're providing biometrics, etc.).
This is a free resource built to help contractors understand what they're signing. After reviewing dozens of AI contractor agreements, patterns emerged that seemed worth documenting publicly.
No platform or company is affiliated with this resource. It's not funded by anyone with a stake in the outcome. Questions or submissions: [email protected]
Context matters. A broad IP clause is more concerning if you're an AI researcher with side projects than if you're doing commodity labeling with no plans to work in the field. A non-compete matters more if you're in California (where they're void) versus Texas.
The Clause Patterns section explains what each clause means in practice. Focus on the ones that intersect with your actual situation and future plans.
This is a common experience reported by contractors. Many complete unpaid training only to find empty task queues for weeks. Platforms typically make no guarantee of minimum hours or work availability - this is usually stated in the agreement.
The contract binds you to restrictions (IP, non-compete, data collection) regardless of whether you receive paid work. Check the "No Guaranteed Work" pattern in the Clause Patterns section.