Scientists Own Their Work

Today we are making a commitment to scientists using Lemma: we won’t front run your work.

By that, we mean that Axiomatic AI does not use your private work to race you to publication or patenting, or train capabilities for others without your permission. Use of Lemma does not give us a right to commercialize or publish your research.

Our goal at Axiomatic AI is to support scientists and engineers by building the systems that they can trust. We are building something different than companies that train on typical human data: verifiable AI for science and engineering, grounded in formal reasoning, physical models, experimental evidence, and tools scientists already use. We win by supporting you on your path to answers.

Your research remains yours

Axiomatic does not claim ownership of your inputs, discoveries, inventions, or publications merely because you use Lemma. We assign to you any rights we may have in outputs generated specifically for you.

We retain our rights in Lemma and in the technology we bring to the work. You give us the limited rights needed to provide, secure, and support the service and to carry out your instructions.

Private work is not product-development material

To implement this, we have three rules about how we behave.

First, we do not use private work to train or fine-tune models, construct cross-user evaluation sets, or improve capabilities for other users without the user’s explicit opt-in. Private work includes prompts, documents, notebooks, code, data, conversations, tool traces, workflows, and results.

Second, our own team will not use the information we need to provide support or operate the service to advance an Axiomatic publication, patent, or independent research result, except as required by law. Access is limited to legitimate support, security, reliability, and legal needs, including enforcement of our terms.

Third, we do not turn a private workflow into a method for someone else. Removing a scientist’s name does not make the scientist’s idea ours. Combining several private traces does not change that. Lemma will not synthesize shared skills, templates, or recommendations from private workflows without the user’s deliberate authorization.

For now, that means the rule is simple: shared capabilities come from public, licensed, or synthetic material, not from private user work.

Contribution options are clear and specific

As a scientist, you might want to contribute to the field by sharing new benchmarks, methodologies or tools. We want to make that possible, while always giving you control over the what, when, and how. We are working on implementing this in the backend. Until those controls are available, we will not treat private user material as a contribution to model training, cross-user evaluation, or a skill marketplace.

Finally, our expectation and hope is that Lemma will provide meaningful contributions to your work. These contributions will be considered part of your work, and we will not use them to extract learnings for our models.

We hope these positions resonate with you and ensure that Lemma remains the platform of choice for scientific exploration, evaluation, and communication. We’re committed to working directly with scientists to further improve it. Lemma is currently in open beta. If you want to help shape the future of AI-assisted scientific exploration, you can sign up via lemma.axiomatic.ai. We’re looking forward to hearing your feedback.