The rules for using AI Chat Console

These Terms & Conditions ("Terms") govern your access to and use of AI Chat Console (the "Service"), a self-hosted platform that routes your prompts to third-party AI providers and models on your behalf and returns their output to you. By creating an account, using a guest/trial session, or otherwise using the Service, you agree to these Terms. If you do not agree, do not use the Service.

Last updated: September 29, 2026. See also our Privacy Policy and Fair Usage Policy, which are incorporated into these Terms by reference.

Jump to: Our role & the AI value chain · Provider terms pass-through · Liability & indemnity · Bangladesh jurisdiction & compliance · Prohibited content & use · Content moderation & guardrails · Reporting, takedowns & legal escalation · Accounts, credits & billing · Intellectual property & output ownership · Suspension & termination · Regulated topics & Modules · Projects, sharing & collaboration · Browser extension · Disclaimers · Changes to these Terms · Contact

AI Chat Console does not train, own, or operate the underlying AI models available through the Service. We operate as a deployer / intermediary in the AI value chain: we integrate with third-party model providers — currently including OpenAI, Anthropic, Google Gemini, xAI/Grok, Mistral, DeepSeek, Cloudflare Workers AI, NVIDIA NIM, Pruna AI, OpenRouter, and any others enabled by the site administrator — and present a unified chat, document-generation, image, video, and audio interface on top of them.

Under this model:

• We are responsible for how our own platform is built and operated — authentication, credit metering, logging, your account data, and the features described in our Privacy Policy.
• Each underlying provider is independently responsible for the training, behavior, safety systems, and output of its own model. We do not control, and cannot guarantee, what any third-party model generates.
• Where a jurisdiction's AI regulation treats "high-risk" deployment differently from a plain pass-through integration, we position ourselves as a general-purpose chat and generation interface, not a provider of a specialized high-risk system (e.g. medical diagnosis, credit scoring, biometric identification, or law-enforcement risk assessment) — using the Service for those purposes is outside its intended use and is done at your own risk and responsibility.
• We may be subject to compliance audits, regulator inquiries, or platform-level reporting obligations relating to how AI systems are deployed through the Service, and we reserve the right to adjust features, model availability, or these Terms to remain compliant with applicable law.

Your use of any model made available through the Service is also subject to that provider's own acceptable-use policy, license restrictions, and usage terms (e.g. OpenAI's Usage Policies, Anthropic's Usage Policy, Google's Generative AI Prohibited Use Policy, Mistral's terms, xAI's terms, and the equivalent for every other provider integrated). These third-party terms are not authored or controlled by us, may change without notice from the provider, and take precedence over anything in these Terms that would conflict with them for that specific provider's model.

By using any model through the Service, you agree that:

• You will not use a model for any purpose its provider's usage policy prohibits, including but not limited to generating disallowed content categories that provider defines (e.g. CSAM, malware, weapons development instructions, non-consensual intimate imagery, or content designed to deceive or defraud).
• You will not attempt to jailbreak, bypass, or circumvent a model's built-in safety systems, content filters, or usage restrictions — through prompt injection, adversarial prompting, or any other technique — whether the model is accessed through our shared pool keys or your own "bring your own key" (BYOK) credentials.
• You are solely responsible for complying with any export-control law applicable to a given model or provider (for example, U.S. EAR/OFAC restrictions some providers apply to certain countries, technologies, or restricted parties). If you are accessing the Service from, or on behalf of, a sanctioned or export-restricted jurisdiction or party, you must not use the affected provider's models.
• Where you bring your own provider API key (BYOK), your relationship with that provider for billing, rate limits, and policy enforcement is directly between you and the provider — we are not a party to it.

We keep standard request metadata (provider, model, token/asset counts, timestamps) for billing and audit purposes as described in our Privacy Policy, so that downstream obligations to a provider or a regulator can be documented if ever required.

Router / broker providers

Some providers offered here are routers (also called brokers or aggregators) rather than the company that runs the model — OpenRouter and Hugging Face Inference Providers are the current examples. When you select a model through a router, your message travels through two organisations, not one: the router itself, and whichever underlying provider it forwards the request to (which may differ between requests for the same model, and may change without notice).

You should understand, and by using a routed model you accept, that:

• The router's own terms, privacy policy and logging practices apply in addition to those of the underlying provider. Neither is authored or controlled by us.
• Which underlying provider serves a given request is decided by the router, subject to any routing preferences we send on your behalf. We cannot guarantee a specific one.
• Capability, pricing and availability information for routed models is supplied by the router and reflects its own assessment of third parties.

Zero Data Retention (ZDR)

Zero Data Retention is a contractual arrangement between this site's operator and a provider — it is not something an application can switch on by itself. Providers grant it at the organisation/account level, subject to approval, and some exclude specific models from it entirely even under a signed agreement.

Accordingly, ZDR mode is only offered here for providers where the operator has confirmed such an agreement is in place. Where it applies, it covers every model call your chat causes — the reply itself and the background calls behind auto-titling, conversation compression, memory, search-query rewriting and prompt enhancement — and those background calls are restricted to covered models or skipped entirely rather than sent outside the arrangement.

For routed providers, ZDR additionally depends on the router supporting and honouring a zero-retention routing mode; where one exists we request it on every such call. Nothing in this section is a warranty of a third party's conduct: our obligation is to request the controls a provider offers and to restrict which models may be used, not to guarantee the provider's own compliance with its published policy.

Web search, voice input & other non-model services

Turning on web search sends your query outside the privacy arrangements that cover the AI model. When web search is enabled for a message, a search query derived from what you typed is sent to a third-party search provider (such as Tavily, Serper, Brave Search, Exa or Parallel, depending on what this instance is configured to use), and the pages it returns are fetched. Those providers receive the query under their own terms and retention policies, which we do not control and which are separate from, and not covered by, any Zero Data Retention arrangement that applies to the model answering you. The same applies to voice input, where your recording is sent to a third-party transcription provider.

If a conversation must not leave a zero-retention boundary, do not enable web search for it, and do not use voice input. Both are opt-in and off unless you or an administrator turn them on.

• No warranty on model output. AI-generated text, code, images, video, audio, and documents may be inaccurate, biased, outdated, offensive, or simply wrong. We do not review, curate, or guarantee the correctness, legality, or fitness for any purpose of model output, and we make no warranty — express or implied — about it.
• No warranty on availability. The Service depends on third-party providers that we do not control. We do not guarantee any specific model, feature, or provider will be available, error-free, or uninterrupted, and a provider may change, deprecate, or restrict a model at any time without our involvement.
• User responsibility. You are solely and fully responsible for: the prompts and inputs you submit; how you use, publish, or act on generated output; and any consequence, legal or otherwise, of doing so. You must independently verify any factual, legal, medical, financial, or safety-critical output before relying on it.
• Indemnity. To the maximum extent permitted by applicable law, you agree to indemnify and hold harmless AI Chat Console, its operator, and its personnel from any claim, liability, damage, loss, or expense (including reasonable legal fees) arising from: your use of the Service in violation of these Terms, a model provider's usage policy, or applicable law; content you submit or generate; or your violation of any third party's rights.
• Limitation of liability. To the maximum extent permitted by applicable law, AI Chat Console shall not be liable for any indirect, incidental, special, consequential, or punitive damages, or any loss of profits, data, or goodwill, arising from your use of the Service — whether the underlying cause is our platform, a third-party AI provider, or a payment gateway. Nothing in these Terms excludes or limits liability that cannot lawfully be excluded or limited under Bangladesh law or another applicable mandatory law.

Governing law

These Terms are governed by the laws of the People's Republic of Bangladesh, including — where applicable to your use of the Service — the Cyber Security Act, 2023 (and its predecessor, the Digital Security Act, to the extent still relevant to conduct that occurred under it), the Information and Communication Technology Act, 2006 (as amended), and applicable data-protection and consumer-protection directives issued by the Government of Bangladesh or its regulatory bodies (including the Bangladesh Telecommunication Regulatory Commission, where relevant). Any dispute arising from these Terms is subject to the exclusive jurisdiction of the competent courts of Bangladesh, without prejudice to any additional local-law obligations that may apply to you in your own country of residence or use.

Nothing here overrides Bangladesh law

This section is a general compliance framework, not a substitute for a lawyer's review of your specific use case. If you operate the Service commercially, or use it for a regulated purpose, you should obtain independent legal advice on your obligations under Bangladeshi law.

In addition to each model provider's own prohibited-use policy (see above), you must not use the Service, in Bangladesh or in any jurisdiction whose law applies to your use, to generate, request, store, or distribute content that:

• Incites violence, terrorism, or hatred against a person or group, or threatens public order or national security.
• Is defamatory under Bangladesh's criminal or civil defamation law, or otherwise knowingly false and damaging to a person's or entity's reputation.
• Violates the Cyber Security Act, 2023, the ICT Act, 2006 (as amended), or any other Bangladeshi statute — including content that is obscene, that outrages religious sentiment in a manner prohibited by law, or that constitutes cyberbullying, harassment, or stalking of an identifiable person.
• Infringes a third party's intellectual property, privacy, or publicity rights, including generating a real person's likeness, voice, or identity without their consent for deceptive or harmful purposes (e.g. non-consensual deepfakes).
• Facilitates fraud, phishing, malware creation, unauthorized system access, or other cybercrime.
• Sexually exploits or endangers a minor in any way, under any circumstance — this is an absolute prohibition with zero tolerance, reported to the relevant authorities immediately upon detection.
• Violates any applicable export-control, sanctions, or exclusion-list requirement of a model provider (see "Model provider terms pass-through" above).

The Service relies on a layered approach to reduce the risk of prohibited content, matched to what is actually implemented in this platform:

• Provider-level safety systems (primary layer). Every request is first subject to the safety filters and moderation built into the underlying AI provider/model itself (OpenAI, Anthropic, Google, etc.) — these are the same industry safety systems that provider applies to all its customers, and they are the main technical barrier against disallowed content categories.
• Feature-level gating. Higher-risk capabilities (e.g. document/file generation, image, video, and audio generation, web search, and deep research) are opt-in toggles that an admin and/or the user must explicitly enable, and are subject to plan-level access controls and daily fair-use limits — see the Fair Usage Policy.
• Administrative oversight (human layer). Site administrators can review platform activity through an audit log of administrative actions, manage plan/model access, and suspend or terminate an account found to violate these Terms. There is no dedicated 24/7 human moderation team pre-screening every message before it reaches a model — moderation of message content itself is performed by the provider-level safety systems described above, not by AI Chat Console staff reading your conversations in real time.
• Zero Data Retention & encryption options do not weaken this moderation approach — they affect storage and provider retention of your data, not whether the provider's own safety systems evaluate your request before generating a response.

If a future version of the Service adds a dedicated pre- or post-generation moderation model, manual review queue, or user-verification step for high-risk interactions, this section will be updated to describe it accurately — see "Changes to these Terms" below.

If you become aware of content generated or hosted through the Service that you believe is illegal, infringing, or otherwise violates these Terms, contact us using the details below with enough detail to locate the content (chat ID or shared-link URL, date, and a description of the issue).

• We will acknowledge a good-faith report and take appropriate action — which may include removing a shared link, disabling a feature for an account, or suspending an account — within a reasonable time frame appropriate to the severity of the report (illegal content involving a minor, an active threat, or a law-enforcement order is treated as the highest priority and actioned as soon as practicable, ahead of routine reports).
• Government & law-enforcement requests. We will respond to a lawful takedown order, data-disclosure request, or cyber-emergency directive issued by a competent Bangladeshi authority (or another authority with valid legal jurisdiction over the Service or its operator) in accordance with applicable law, and will comply with the timeframe that order specifies where legally required to do so.
• Nothing above is a guarantee of a specific response time absent a legally binding requirement — this describes our intended escalation priority, not a contractual SLA to end users.

• You must provide accurate registration information and keep your login credentials confidential. You are responsible for activity that occurs under your account.
• The Service is credit-metered — chat, document, image, video, and audio generation consume credits and/or count against daily fair-use limits as described in the Fair Usage Policy. Purchased credits and subscription fees are generally non-refundable except where required by applicable consumer-protection law or explicitly stated at time of purchase.
• Payments are processed by a third-party gateway (Stripe, SSLCommerz, bKash, or a manual/offline method) under that gateway's own terms; we do not store your full card number.
• Paid plans bill on a 30-day cycle beginning when payment is confirmed (or a 365-day cycle where a plan is offered and purchased yearly). An invoice is issued at the start of each cycle and is payable by the due date shown on it.
• We do not automatically charge a stored payment method. Each renewal is initiated by you; a subscription that is not renewed does not continue.
• If an invoice is unpaid, access to the paid plan continues for a short grace period, after which the account is moved to the free plan. Moving to the free plan changes what the account may do going forward; it does not delete stored conversations or remove credits already held. See the Billing Guide for the current grace period and the full sequence.
• Guest/trial access (where enabled) is provided without an account, is not guaranteed to persist chat history, and may be limited or withdrawn at any time.
• Generated content is not permanent storage. Images, video, audio, and documents you generate are removed when the chat that contains them is deleted, and unreferenced/orphaned files are periodically cleared by an automated maintenance job — once removed, a file cannot be recovered. Download anything you want to keep. See the Fair Usage Policy for full details before you rely on the Service to hold generated content long-term.

• We claim no ownership over the prompts you submit or the output a model generates for you, subject to the underlying provider's own terms (some providers impose their own conditions on commercial use of their model's output — check the specific provider's terms if this matters to you).
• Because outputs are AI-generated, their copyright status can be legally uncertain in some jurisdictions and may depend on the level of human creative input — we make no representation about whether a specific output is protectable or freely usable by you.
• You must not submit content that infringes a third party's intellectual property rights, and you are responsible for any resulting claim (see "Liability & indemnity" above).
• The AI Chat Console platform itself — its software, branding, and design — remains the property of its operator and is not licensed to you beyond what's needed to use the Service as intended.

We may suspend or terminate your access to the Service, with or without notice, if we determine in good faith that you have violated these Terms, a model provider's usage policy, applicable Bangladeshi or other law, or if required to do so by a competent authority. You may delete your own account at any time from Account settings. Sections of these Terms that by their nature should survive termination (including Liability & indemnity, Bangladesh jurisdiction, and Intellectual property) continue to apply after your account is closed.

Some Modules — the guided, form-based tools in the app — draft material in areas where a wrong answer has real consequences: legal, financial and tax, health, and employment. These Modules are labelled in the app, carry a notice before you use them, and their output carries a caveat line. This section explains what that labelling means and what it does not.

What you agree to when you use them

• Review before reliance. You will have output from these Modules reviewed by a person qualified in the relevant field and jurisdiction before you sign it, send it, file it, publish it, or otherwise act on it.
• Verify the facts. AI models produce plausible text, which is not the same as correct text. Dates, figures, citations, statutory references, clause numbers, qualifications and employment history may be wrong or invented. Checking them is your responsibility.
• Jurisdiction is yours to supply. The model works only from what you tell it. Unless you state a country and, where relevant, a state or division, it may apply assumptions that do not hold where you are. Bangladeshi law in particular is under-represented in the training data of every general-purpose model on the market.
• Not for emergencies. Nothing in the Service is suitable for an urgent medical, safety or legal situation. Contact the appropriate emergency service or a qualified professional directly.
• Decisions about people. Where a Module summarises or screens candidates, employees or applicants, its output is a shortcut for your reading, not an assessment. You remain the decision-maker and remain responsible for the fairness and lawfulness of that decision, including under any applicable anti-discrimination and labour law.

What we do

• Modules in these categories are marked in the gallery, and show a notice naming the category before you fill anything in.
• Their instructions to the model require it to state its assumptions and show its reasoning, so a professional can check the work rather than having to redo it, and to mark missing details as placeholders instead of inventing them.
• Their output ends with a caveat line, so the warning stays attached to the text when it is copied, exported or forwarded.
• These measures are safeguards, not guarantees. They reduce the chance of a confident wrong answer; they do not remove it, and they do not transfer responsibility for the outcome to us.

Modules created by other users

Where the Service allows Modules authored or shared by other users, those Modules are user-generated content. We do not review or endorse them, they may carry no professional oversight whatsoever, and the labelling above may not have been applied correctly by their author. Treat output from a user-authored Module with more caution than our own, not less. You can report one to us at any time using the contact details below.

Nothing in this section limits any right you have under the consumer protection law of Bangladesh or any other mandatory law that applies to you.

Projects. A project is a workspace holding instructions, notes, memory and uploaded files that are automatically supplied to every chat inside it. You are responsible for what you put in one. Do not upload material you do not have the right to use, and do not place another person’s personal data in a project’s knowledge base without a lawful basis for doing so.

Uploaded knowledge is stored as readable text so it can be searched. A project’s instructions, memory and knowledge files are not covered by end-to-end chat encryption, which protects message content only. Keep anything requiring that protection out of a project.

Sharing a conversation. You may share a chat as a read-only link. Anyone holding that link can read the conversation, so treat it as public: do not share a chat containing confidential or personal information you are not entitled to disclose. You can revoke a share link at any time, though we cannot recall anything already copied from it.

Collaborators. Inviting someone into a chat or project gives them access to its contents and, where you grant it, the ability to send messages within it. Usage by a collaborator is billed to the account that owns the conversation — yours — and counts against your plan limits. You remain responsible under these Terms for activity conducted in workspaces you own, so invite only people you trust and remove access when it is no longer needed.

Removing a collaborator stops future access; it does not retrieve information they already saw.

Memory. The Service can retain facts drawn from your conversations in order to use them in later ones. This is under your control: memory can be disabled for your whole account, set off by default for new conversations, or switched off for any single conversation, in which case nothing is recalled into that chat and nothing is learned from it. Stored facts are listed in your account and can be deleted individually at any time. Facts learned inside a project are held against that project and are not used elsewhere.

The browser extension is part of the Service and these Terms apply to it in full. It operates on the same account, the same credit balance and the same limits as the website — it is not a separate product or a separate allowance.

Page content is sent only when you send it. The extension does not read pages in the background. Content is transmitted when you explicitly capture the current page, another tab or a screenshot, and it is then handled exactly like anything else you type into a message, including being passed to the model provider you selected.

You are responsible for what you capture. Do not send us content from pages you are not permitted to share — another party’s confidential systems, material behind an agreement that forbids it, or personal data you have no lawful basis to process.

Access tokens. The extension authenticates with a revocable token rather than your password. Keep it to yourself, and revoke it from your account if a device is lost or you no longer use the extension. Activity performed with a valid token is treated as your activity.

We may update these Terms as the Service, its provider integrations, or applicable law evolve (for example, adding a new AI provider, a new regulated feature, or a new Bangladeshi compliance requirement). The "Last updated" date above reflects the most recent revision. Continued use of the Service after a change constitutes your acceptance of the updated Terms; if you do not agree with a change, you should stop using the Service and may delete your account.

Questions about these Terms, a takedown request, or a legal/compliance inquiry can be sent to support@aichatconsole.com.