Privacy policy

Effective date: July 19, 2026

The waitlist has been collecting entries since July 10, 2026, when the site went live; this policy applies to all entries, including those collected before it was published.

This policy covers naomilm.com and its waitlist.

What we collect. The waitlist form asks for your email address and, optionally, your name. Nothing else is asked of you.

We measure page visits with GoatCounter, an open-source analytics tool running on our own server. It sets no cookies, stores no IP addresses, and collects nothing that identifies you. No data leaves our infrastructure and nothing is shared with any third party. That is the entire extent of measurement on this site.

Why. One purpose: to write to you when NaomiLM opens access, with the founding-member offer we promised. No newsletter, no advertising. We said one honest email; this policy is bound by that.

We will never sell your data. Not to anyone, not ever.

Who processes it. Your submission is stored in Google Drive; Google acts as our processor, only on our instructions. Nobody else gets it unless the law compels us.

How long. Until access opens and you have had a fair chance to join. Unconverted entries are deleted after that, and you can be deleted at any time: one email to privacy@naomilm.com removes you and your data.

Your rights. Wherever you are in the world, ask privacy@naomilm.com to see, correct, or delete what we hold. We honor these requests for everyone, not only where a law requires it. EU/EEA residents hold these rights under the GDPR; Israeli residents under the Privacy Protection Law, 5741-1981.

When the app launches, it will handle data this waitlist does not (your conversations with the model). A fuller policy will replace this one before then, and the access email will link to it.

Conversation data retention. Conversations with NaomiLM are logged server-side for safety monitoring and legal defense. Conversation logs are automatically purged after 180 days unless a litigation hold is active. You may request deletion of your conversation data at any time via privacy@naomilm.com; requests are fulfilled within 30 days unless the data is subject to a litigation hold. Hashed identifiers (IP hash, session hash) are stored; raw IP addresses are not retained.

Training and your conversations. By default, your conversations are not used to train NaomiLM. They are stored so you can return to them, and that is all.

If you would like to help improve the model, there is a switch in your settings called Help improve NaomiLM. It is off until you turn it on. When it is on, your conversations and your thumbs-up or thumbs-down ratings may be reviewed and used to train future versions of NaomiLM. Before review, they are stripped of names, email addresses, and other identifying details. They are never sold, never shared with third parties, and never used to train anything other than NaomiLM.

You can turn the switch off at any time. From that moment, new conversations stay out of training. If you delete your account, any of your data that has not yet been used for training is removed from the training pool along with everything else.

One thing we say plainly because most companies do not: once a conversation has been used to train a released version of the model, it cannot be extracted or removed from that version's weights. Turning the switch off, or deleting your account, governs the future. It cannot rewind a model that has already been trained.

Not directed at children under 13. If this policy changes, the new version lives at this address with a new date.

Terms of use

Effective date: July 19, 2026

By using NaomiLM (the model, chat interface, or any service at naomilm.com), you agree to these terms.

What NaomiLM is. A language model fine-tuned against sycophancy. It is not therapy, not medical advice, not legal counsel. It is trained to be honest, not to be correct about specialized domains. Do not rely on it for decisions that require licensed expertise.

Age requirement. You must be at least 13 years old to use NaomiLM. If you are between 13 and 17, you may use NaomiLM but you acknowledge that it is not a substitute for parental guidance, medical advice, or licensed professional support. NaomiLM will not assist minors with obtaining medication, circumventing parental authority, or any action that poses a safety risk to a person under 18.

Prohibited uses. You may not use NaomiLM to:

Obtain instructions for weapons, explosives, chemical or biological agents, or 3D-printed firearms.
Obtain methods of self-harm, suicide, or harm to others.
Generate content that sexualizes minors.
Impersonate another person or misrepresent AI-generated text as human-written in regulated contexts.
Systematically extract training data or reverse-engineer model weights.
Circumvent safety measures for the purpose of documenting failures for litigation.
Generate, curate, or amplify content intended for propaganda, harassment, or political manipulation.
Use model outputs to dehumanize, target, or incite violence against any group or individual.

Good-faith reporting obligation. NaomiLM publishes its principles and provides a reporting mechanism for violations. If you discover behavior that violates a published principle:

You agree to report it through our mechanism within 60 days of discovery, before pursuing any legal claim related to that behavior. Failure to report within this window extinguishes the claim.
Deliberate reproduction of a known violation, without reporting, for the purpose of litigation constitutes bad faith and a breach of these terms.
Jailbreaking, prompt injection, or adversarial manipulation to force a violation, then claiming the model produced it unprompted, constitutes fraud.
Screenshots or excerpts of model responses shared publicly without the full conversation context (including your prompt) constitute misrepresentation if used to support a legal or regulatory claim.

Responsible disclosure for researchers. Security researchers and academics who discover violations are welcome. We ask for 90 days notice before publication, submitted through the reporting form or to safety@naomilm.com. Researchers who follow responsible disclosure will be credited and will never face legal action from us.

Honest responses are not harm. NaomiLM is designed to disagree with you, challenge weak reasoning, and decline to validate poor decisions. A response that is unflattering, direct, or that contradicts your expectations is the product working as intended. Emotional discomfort from honest feedback is not grounds for a claim of harm.

User responsibility. You are responsible for how you use NaomiLM and its outputs. Intentional misuse, including deliberate attempts to obtain harmful content, constitutes a breach of these terms. The prohibited uses listed above are examples, not an exhaustive list. Any use of NaomiLM intended to cause harm to yourself, another person, or a group, whether or not specifically listed, falls under the no-harm principle and constitutes a breach of these terms. NaomiLM is not liable for harm resulting from prohibited use.

No warranty. NaomiLM is provided as-is. We do not guarantee accuracy, completeness, or fitness for any particular purpose. The model will sometimes be wrong. When it is wrong, report it. Model outputs are generated algorithmically and do not represent the views, opinions, or factual assertions of NaomiLM's operator, its founders, or its employees. NaomiLM may produce inaccurate statements about real people, organizations, or events. These are model errors, not editorial positions. The operator of NaomiLM is not liable for defamation claims arising from algorithmically generated content.

Termination. We may suspend or terminate access for any violation of these terms. You may stop using NaomiLM at any time. Data deletion is available per the privacy policy above.

Litigation hold. If you initiate or threaten legal proceedings against NaomiLM or its operator, your right to data deletion is suspended and your account access may be restricted for the duration of the dispute. Data relevant to the claim will be preserved as required by law. This hold remains in effect until the matter is fully resolved. Deleting your account or data before or during a dispute you initiated may be treated as spoliation of evidence and bad faith.

Dispute resolution. Before initiating any legal proceeding, you agree to attempt resolution through good-faith mediation. Either party may propose a mediator; if no agreement on a mediator is reached within 30 days, the Israel Bar Association will appoint one. No lawsuit may be filed until mediation has been attempted and failed or 60 days have passed since the mediation request, whichever comes first.

Governing law. These terms are governed by Israeli law. Disputes are subject to the exclusive jurisdiction of the courts in Tel Aviv-Yafo.