Data transparency
The MMF Behavioral Institute is funded by licensing behavioral video that families volunteer, under permissions they grant one at a time, in our pay-for-data program. This page is the whole story, including the parts that do not flatter us.
Before anything is collected you read the grants one at a time and answer each one separately. There are 5 of them, listed below in full. Joining the program does not grant all 5. The consent document says, for each grant, what it covers, who can license the footage under it, what you are paid, and what withdrawal actually does. The framework is still with counsel and will be published here before the program opens.
You record everyday footage at home and upload it through a private portal. Nothing is posted publicly, nothing is pulled off your devices, and the free services collect no footage in the background.
Trained reviewers check each contribution against the usable-footage criteria and flag anything that should not enter a dataset. Whether that footage earns you anything depends on which grants you made. Payment is attached to the two commercial grants: commercial use (C2) and the Institute’s own technology (C3). Granting either one puts you in the paid program at $10 per 15 minutes of usable footage. Research (B) and nonprofit tools (C1) are unpaid. Universities and nonprofits do pay us a licence fee for those, but their licence bars them from selling anything built on your child’s footage, so there is no sale to share with you. Commercial research sits in C2, not B, so nobody can buy commercial access at an unpaid rate.
Names, spoken identifying details, addresses, school insignia, and similar markers are removed or obscured under a published standard before footage enters any dataset. What that does and does not accomplish for video is the last section on this page, and it is worth reading before you decide anything.
Datasets go to vetted organizations under the specific grants you made and no further. A family who granted research only is never in a dataset licensed for technology development. Licensees are contractually restricted: no attempts at re-identification, no resale, no sublicensing, no use outside the licensed purpose. Research and nonprofit licensees carry one more restriction, which is that they may not sell anything they build on the footage.
Licence fees pay for the resource library, the courses, the video observation service, and your compensation. Universities and nonprofits pay those fees too, which is why the free side can exist without ads. What separates their licence from a commercial one is that they are barred from selling anything built on the footage. That is the entire business model; there is no hidden second one.
Consent here is not one switch. There are 5 separate grants, listed in order of how far a video travels from your living room. The first is part of the service itself. The other 4 are opt-ins you grant or refuse one at a time, in any combination you like. Refusing all 4 is an ordinary choice, and it changes nothing about the resource library or the courses. Each card below carries a chip saying whether that grant earns you money, and the reasoning under it. Paid for commercial use. Research and nonprofit use are unpaid contributions.
This one is how the service works at all. If you send us footage and ask a BCBA what they see in it, somebody here has to watch it. You cannot have the observation without granting the viewing.
Using your footage to do the thing you asked us to do.
“The Institute may review the footage I submit in order to give me feedback, check its quality, and improve its own review standards.”
What this grant permits
Research is a wider circle. Your footage leaves our hands under license and goes to people studying how children develop and behave. Their work gets published, and published work does not come back.
Research into child development, autism, ADHD, and behavior.
“My child’s footage, with identifying details removed and my family referred to by a code rather than a name, may be included in datasets licensed to universities and nonprofit or public research institutions for research into child development and behavior. Those institutions pay the Institute a licence fee and are barred from selling anything built on the footage. I understand I am not paid for this grant.”
Unpaid. Universities and nonprofit institutions do pay us a licence fee, and we are not hiding that. What they cannot do is sell a product built on your child’s footage; their licence forbids it, and their output is published research. Because nobody at the end of this chain is selling anything, there is no sale to share with you. If one of those institutions later wants to commercialise, they need a commercial licence instead, and then you are paid.
What this grant permits
The last three cover what gets built on the footage, split by who ends up owning it. C1 is universities and nonprofits. C2 is companies, and it covers both the products they sell and the research they run to get there, so a business cannot buy access at the unpaid research rate. C3 is us, training our own models and selling the result to pay for the free programs. That last one is the reason this Institute can exist, which is exactly why it gets its own line and its own refusal instead of hiding inside the other two.
Universities and nonprofits building tools for children like yours.
“My child’s coded footage may be used by universities and nonprofit organizations to build observation, teaching, and assistive tools for children with developmental differences, including software trained on that footage.”
Unpaid, on the same terms as the research grant. Nonprofit and academic developers pay us a licence fee and are barred from selling what they build on your child’s footage. If they want to sell it, they need a commercial licence, and then you are paid.
What this grant permits
Companies doing research or building products they sell, for this same population.
“My child’s coded footage may be licensed to companies, both for their own research and to build behavioral observation and support products they sell for profit, limited to work concerning children with developmental differences.”
Paid. A company profits here, whether it sells a product or runs the research to build one, so the family whose child is in the footage is paid too. Research done by a company sits in this grant rather than the research grant, precisely so that nobody can buy commercial access at an unpaid rate.
What this grant permits
The Institute training its own behavior-detection software.
“My child’s footage may be used by the Institute to train its own computer-vision and machine-learning models for observing and measuring the behavior of children with developmental differences, including products the Institute sells.”
Paid. We sell what we build here, and it is the revenue that keeps the library and the courses free and pays for the reviewers who write observations. Since we profit, you are paid.
What this grant permits
Payment tracks profit. Somebody buys a licence and profits from your child's footage, either by selling what they build or by running the research that gets them there, and the family in that footage gets a share. That is the whole test, and it is why commercial use (C2) and our own technology (C3) are the two grants that pay. Here is the reasoning for each grant, in the same words the consent documents use.
Behavioral research (B)Unpaid. Universities and nonprofit institutions do pay us a licence fee, and we are not hiding that. What they cannot do is sell a product built on your child’s footage; their licence forbids it, and their output is published research. Because nobody at the end of this chain is selling anything, there is no sale to share with you. If one of those institutions later wants to commercialise, they need a commercial licence instead, and then you are paid.
Nonprofit tools (C1)Unpaid, on the same terms as the research grant. Nonprofit and academic developers pay us a licence fee and are barred from selling what they build on your child’s footage. If they want to sell it, they need a commercial licence, and then you are paid.
Commercial use (C2)Paid. A company profits here, whether it sells a product or runs the research to build one, so the family whose child is in the footage is paid too. Research done by a company sits in this grant rather than the research grant, precisely so that nobody can buy commercial access at an unpaid rate.
Our own technology (C3)Paid. We sell what we build here, and it is the revenue that keeps the library and the courses free and pays for the reviewers who write observations. Since we profit, you are paid.
On the other side of the line sit behavioral research (B) and nonprofit tools (C1). Unpaid there does not mean unvalued. A family that grants research and refuses everything else has given something real, and the studies that come out of it are part of why this program exists. We would rather tell you that the licence fee exists and that the grant still earns you nothing than hide the fee or attach a token payment to make this page look generous.
So you can contribute to research alone and be paid nothing, on purpose, with your eyes open. You can refuse both commercial grants and stay in the program: refusing them does not gate enrollment, your place in a course, or a single thing in the free library. The one service that does turn on a grant is video observation, and the grant it asks for is the research one rather than a commercial one. That exchange is described on its own page, and we would rather name the condition than keep calling the service unconditionally free.
Each of the 4 opt-ins is refused on its own. You can grant research and refuse every technology grant. You can grant the nonprofit one and refuse the two commercial ones, ours included. Whatever you decide, our pay-for-data page shows the same split before you answer anything.
These are uses we will not make and will not license, whatever the contract on the table is worth. They are written down so you can hold us to them later.
Contributed footage is never used to advertise anything, including our own programs, and never appears in marketing material.
We do not build or supply data for models that recognize human behavior generally. Every model trained on this data is scoped to children with developmental differences.
No workplace monitoring, no crowd or pedestrian analysis, no security or retail surveillance, no fall-detection or elder monitoring systems.
Contributed footage is never used to study shopping behavior or test commercial products on children.
We do not license to these uses and will contest any demand that is not backed by valid legal process.
Licensees are contractually barred from attempting to re-identify any child or family, and from reselling or sublicensing what they receive.
The pay-for-data program will not open in 7 states. Each has a statute that conflicts with paying a family for behavioral video and then licensing that video onward. In some, the ban on profiting from biometric identifiers is flat, so no amount of consent cures it. In others, selling a minor's data or sensitive health data is prohibited outright. Where the language is ambiguous we exclude the state instead of finding out in court what it meant.
| State | Law | Why it rules the program out |
|---|---|---|
| ILIllinois | Biometric Information Privacy Act (740 ILCS 14), § 15(c) | Bars a private entity from selling, leasing, trading, or otherwise profiting from a person’s biometric identifiers. The bar is absolute, so consent cannot cure it. |
| TXTexas | Capture or Use of Biometric Identifier Act (Tex. Bus. & Com. Code § 503.001) | Prohibits the sale, lease, or disclosure of biometric identifiers, including records of face geometry, outside a few narrow exceptions. |
| WAWashington | My Health My Data Act (Ch. 19.373 RCW) and the biometric statute (HB 1493) | Treats behavioral and health information as consumer health data, requires a separate signed authorization before any sale, and lets consumers sue directly. |
| COColorado | Colorado Privacy Act as amended by HB 24-1130 | Prohibits selling, leasing, or trading biometric identifiers. |
| CTConnecticut | Connecticut Data Privacy Act as amended by SB 1295 (effective July 1, 2026) | Prohibits selling the personal data of anyone known to be under 18, and treats disability data as sensitive. |
| MDMaryland | Maryland Online Data Privacy Act | Bans the sale of sensitive personal data outright, and bars selling the data of anyone under 18 even with consent. |
| OROregon | Oregon Consumer Privacy Act as amended | Prohibits selling the personal data of a consumer known to be under 16. |
This exclusion covers payment and licensing, and nothing else. The resource library, every course, and video observation are available in all fifty states, including these 7. If your state is on the list and the law changes, or counsel reads it differently than we do, we will say so here and say when.
Video of a child's face identifies that child. There is no processing step that undoes that, and we are not going to pretend there is. Any program that calls face-on footage anonymous is either confused about the word or using it on purpose.
What we actually do is narrower, and we would rather describe it accurately than sell you a word. Before footage enters a dataset we remove or obscure names, spoken identifying details, addresses, school insignia, and similar markers. Files carry a code instead of your family's name, and the key that connects the code to you stays with us. That is coding, not anonymization. The link back to your family still exists; we hold it, and we control who can use it.
The rest is handled by contract and by access control. Licensees are barred from attempting to re-identify any child or family, from reselling, and from sublicensing, and they receive only the footage covered by the grants families actually made. Contracts are enforcement after the fact. They make re-identification a breach we can act on. They do not make it impossible, and you should read every promise on this page with that in mind.
One limit has no workaround at all. A model that has already been trained on footage cannot have that footage taken back out of it. Training changes the weights, and there is no undo button. This is why the three technology grants are disclosed as permanent for training already done, and why withdrawing from them stops future use rather than reaching backward. If that trade is not one you want to make, refuse those grants. Everything else in the program still works.