01Scope
This Policy applies to the NeuralAI website (neuralai.wasmer.app), the live NEA1 chat (hosted on a workers.dev address and embedded in the website), and the inquiry form (collectively, the “Services”), operated by NeuralAI, an independent AI research lab based in West Memphis, Arkansas, United States, founded by DeAndrew Preston Harris. For purposes of privacy law, NeuralAI is the controller (or “business”) of the personal information described here.
This Policy does not apply to third-party websites or platforms you reach through links from the Services — including Hugging Face, GitHub, and any page the model fetches during a chat. Those services have their own privacy policies, and when you are on their properties, their policies govern. It also does not apply to information contained in the NEA1 training corpora or in the openly released model artifacts themselves, which are governed by the applicable open licenses and dataset documentation.
02What we collect
We collect only the categories of information that the Services genuinely involve. The table below is the complete picture as of the effective date.
| Category | What it includes | Where it comes from |
|---|---|---|
| Inquiry information | Your name, email address, organization (if provided), topic of interest, and the message you write. | The “Work with the lab” inquiry form on the website. Submissions are stored in our lead-management datastore so we can respond and keep track of the conversation. |
| Chat content | The messages you send to NEA1, the model’s replies, the conversation context your browser sends with each new message so follow-up questions work, and the decoding mode you select. | The live chat on the website or the chat page. Your conversation is held in your browser’s page memory while you chat; each request carries recent turns to the server so the model can answer in context. |
| Tool-related content | URLs the model is asked (or decides) to fetch, the text retrieved from those pages, file names involved in demonstration tool calls, and tool results and error messages. | Generated during a chat when a tool is invoked. Retrieved page text is processed to ground the model’s answer. |
| Technical and usage data | IP address, browser and device type, user-agent string, request timestamps, pages and endpoints requested, response status, latency, and error information. | Automatically, from the hosting, edge, and inference providers that deliver the Services (see Section 5). Used for delivery, security, and reliability. |
| Communications | Messages you send us outside the form (for example, replies to an email we sent you), and records of privacy or support requests. | Directly from you. |
| Cookies and similar tech | As of the effective date, these pages do not set advertising, analytics, or social-media cookies. Infrastructure providers may use strictly necessary mechanisms (such as security or load-balancing cookies) — see Section 12. | Your browser and our infrastructure providers. |
What we do not collect. The Services have no user accounts, no payment processing, and no in-product purchases, so we do not collect account credentials, payment-card data, or purchase histories through them. We do not knowingly collect precise geolocation, biometric identifiers, or government ID numbers, and we ask you not to send us sensitive information of any kind through the chat or the form.
03How we use information
We use the information described above for the following purposes, and only those purposes (or purposes compatible with them), unless we tell you otherwise and obtain any consent the law requires:
- To provide the Services — serving the website, generating chat responses, maintaining conversation context across turns, executing tool calls you request, and returning results to your browser;
- To respond to you — answering inquiries, following up on partnership or access requests, and maintaining records of our communications;
- To keep the Services secure and available — detecting abuse, spam (the inquiry form uses a hidden anti-spam field), automated overload, and attacks; enforcing rate limits; and debugging outages and errors;
- To measure reliability — understanding latency, failure rates, and capacity so we can keep the demo running and plan infrastructure;
- To evaluate and improve model quality — reviewing interaction patterns in aggregated or de-identified form to find failure modes (for example, questions the model mishandles) and to design evaluations and future training data that address them;
- To comply with the law — meeting legal obligations, responding to lawful requests, and protecting the rights, property, and safety of NeuralAI, our visitors, and the public.
We do not use personal information for targeted advertising, and we do not build advertising profiles of visitors.
04AI and chat practices
A few practices are specific to talking with a language model, and we want them to be explicit:
- Processing. When you send a message, your message and recent conversation context are transmitted over HTTPS to our chat service, which runs the NEA1 model on inference infrastructure operated for us by a compute provider (currently Modal). The model’s reply is streamed back to your browser token by token.
- Tool calls. If the model calls a tool, our servers may contact the website the model selects and process the returned text. That means a URL you mention — or a variation of it chosen by the model — may be requested by our infrastructure, and the operator of that site may see a request from our servers.
- Don’t send sensitive information. NEA1 is an experimental research system. Do not send passwords, one-time codes, private keys, payment or bank details, medical or benefits information, government identifiers, or personal information about other people. Anything you type into the chat should be treated as content you are comfortable having processed by automated systems and, where Section 3 applies, reviewed in the limited ways described there.
- Review. Quality and safety evaluation is mostly automated and aggregate. When you report a problem, when an inquiry requires it, or when we investigate abuse or a specific failure, authorized lab personnel may review the relevant messages and technical records.
- Output is not confidential output. The model can repeat or transform text it is given. Do not rely on the chat to keep a secret inside the conversation, and be aware that similar outputs may be produced for other users.
06Retention
We keep personal information only as long as reasonably necessary for the purposes described in this Policy. In practice: inquiry-form submissions and related correspondence are kept while we are evaluating or maintaining the relationship they concern, and for a reasonable period afterward for records, follow-up, and dispute purposes; chat content is processed to answer you and is not assembled into a permanent per-visitor profile — transient technical and security logs created by our providers are kept for the limited periods those systems retain them and that security and debugging require; and records connected to abuse or legal matters may be kept longer where needed. When information is no longer needed, we delete it or de-identify it, though residual copies may persist in provider backups for a limited time before being overwritten.
If you would like information you submitted to be deleted sooner, see Section 8.
07Security
We use safeguards appropriate to a small research operation: the Services are served over HTTPS; access to administrative systems and stored submissions is restricted to authorized personnel; credentials and API tokens for our infrastructure are held in access-controlled storage and are never embedded in the public pages; and the chat’s server-side tools are deliberately limited (for example, fetched content is truncated, and file access is confined to a restricted demonstration sandbox). No method of transmission or storage is perfectly secure, however, and we cannot guarantee absolute security. If you believe you have found a security issue in the Services, please report it through the inquiry form with “Security” in your message, and give us a reasonable chance to address it before public disclosure.
08Your choices and rights
Depending on where you live, you may have rights regarding your personal information, which can include the rights to: know what we hold about you and receive a copy (access/portability); correct inaccurate information; delete information we hold about you; restrict or object to certain processing; withdraw consent where processing is based on consent; and not be discriminated against for exercising these rights.
How to make a request. Submit a request through the inquiry form on the homepage with “Privacy request” in your message, and tell us what you are asking for (for example, deletion of an inquiry you sent). Because the chat has no accounts, chat-related requests need enough detail (such as the approximate date and the content involved) for us to locate any records that exist. We may need to verify your identity proportionately to the request — for example, by confirming control of the email address used in an inquiry. We will respond within the timeframe required by applicable law (generally 45 days for U.S. state privacy laws, and one month under the GDPR, extendable as those laws allow). If we decline a request, we will explain why, and you may appeal by replying to our response. You may also use an authorized agent where the law permits, subject to verification.
Your practical choices. You can stop using the chat at any time; closing the page ends the session, and starting a new chat discards the context held in the page. You can decline to use the inquiry form — it is the only feature that asks for your name and email. Browser settings let you block cookies generally; as noted in Section 12, these pages do not depend on tracking cookies.
09U.S. state privacy notices
For residents of states with comprehensive privacy laws (including California, Colorado, Connecticut, Virginia, Utah, Texas, and others as they take effect), the following supplements the rest of this Policy. The categories of personal information we collect are: identifiers (such as name, email address, and IP address), internet or network activity (such as requests and usage logs), the content of communications you send us or the model, and — only if you choose to include it in a message — organization and professional information. The sources, purposes, and disclosure categories for this information are described in Sections 2, 3, and 5.
No sale or sharing. We do not “sell” personal information and do not “share” it for cross-context behavioral advertising, as those terms are defined in state law, and we do not engage in targeted advertising or in profiling that produces legal or similarly significant effects. We do not knowingly sell or share the personal information of consumers under 16. We do not use or disclose sensitive personal information for purposes other than those permitted without a right to limit, and we ask you not to provide sensitive personal information at all. Your rights and how to exercise them — including the right to appeal a declined request — are described in Section 8. We do not recognize browser-based opt-out signals as opt-outs from sale or sharing for the simple reason that we do not sell or share personal information in the first place.
10International visitors
NeuralAI is based in the United States, and the Services are operated from the United States using providers whose infrastructure may be located in the United States and other countries. If you access the Services from outside the United States, your information will be transferred to and processed in the United States and those other locations, whose privacy laws may differ from your own. Where required, we rely on appropriate legal bases for processing — performing the service you request, our legitimate interests in operating and securing a research demonstration, compliance with legal obligations, and your consent where we ask for it. If you are in the EEA, UK, or Switzerland, the rights described in Section 8 apply to you under your local law, and you also have the right to lodge a complaint with your local data-protection authority.
11Children
The Services are not directed to children under 13, and we do not knowingly collect personal information from children under 13. If you believe a child under 13 has provided us personal information (for example, through the inquiry form), contact us through the form with “Privacy request — child data” in the message, and we will delete it promptly. Users under 18 should use the Services only with the consent and supervision of a parent or legal guardian, as stated in the Terms of Service.
13Third-party links and downloads
The Services link to platforms we do not operate — including Hugging Face (weights and model card), GitHub (code and evaluations), and the live chat’s hosting address. NEA1’s weights are distributed through Hugging Face under the Apache 2.0 license; when you download or interact with artifacts there, Hugging Face’s privacy policy governs that activity. We are not responsible for the privacy practices of third parties, and we encourage you to read their policies.
14Changes to this Policy
We may update this Policy as the Services and our practices evolve. The revised Policy will be posted on this page with an updated “Last updated” date, and material changes will take effect when posted unless we state a later date. For changes that materially expand how we use information already collected, we will make reasonable efforts to highlight the change on the website. Your continued use of the Services after an update takes effect means you acknowledge the revised Policy.
15Contact
Privacy questions and requests — including access, correction, deletion, and appeals — can be sent through the inquiry form on the NeuralAI homepage. Put “Privacy request” in your message so it reaches the right handling quickly. NeuralAI is based in West Memphis, Arkansas, United States.
The Terms of Service govern use of the site and the chat, including the research-model notice and the acceptable-use rules.