What the platform is not for.

This policy applies to everyone who uses Transglot: account holders, organization members, contractors working under a seat, and anything automated through the REST API, the command line interface, a webhook or a connector. It is incorporated into the Terms of Service, so a breach of it is a breach of that agreement. It applies to content you store, content you submit for translation, content the Service produces for you and content you deliver through the Service. We update it as the product and the threat landscape change, and the effective date at the top says when we last did.

The list below is illustrative rather than exhaustive. The rule it comes from is simple: do not use the Service to do something unlawful, to harm somebody, or to take from other customers the capacity they are paying for. If a use is not on the list but obviously breaks that rule, it breaks this policy. If a use is not on the list and you are unsure, write to us before you build on it, and we would much rather answer that question than suspend an account over it.

Do not store, submit, translate, generate or deliver content that is unlawful in a jurisdiction where it will be published, that infringes copyright, trademark, patent, trade secret, database, publicity or privacy rights, or that you do not have the rights to submit. Localizing a product into forty languages multiplies the reach of whatever is in the string, which is exactly why the source has to be clean. Copyright complaints are handled under the Copyright and DMCA policy.

Do not use the Service in connection with child sexual abuse material or any sexualisation of minors, in any form, including text, fiction, description, prompt or image. This is the one line with no notice, no cure period and no discussion: we terminate immediately, preserve the evidence, and report to the appropriate authorities and reporting bodies as the law requires or permits.

Do not use the Service to produce or distribute content that incites or threatens violence, that promotes terrorism or violent extremism, that harasses, bullies, stalks or targets an individual, or that attacks people on the basis of race, ethnicity, national origin, religion, caste, sex, gender, gender identity, sexual orientation, disability or serious disease. Translating research, journalism, moderation training data or historical material that quotes such content is not a breach of this clause; producing it for distribution is.

Do not use the Service to localize scams, phishing pages, fake stores, fraudulent investment offers, pirated software or counterfeit goods listings. Do not use it to generate impersonations of a real person or organization, fake reviews, fabricated news, astroturfed political messaging, or coordinated inauthentic content in many languages at once. A translation engine is a force multiplier for deception, and multiplying deception is a breach of this policy even where a single instance would not be unlawful.

Machine translation output is not reviewed by a qualified human unless you review it. Do not rely on unreviewed output for medical instructions, drug labelling, safety warnings, legal notices, regulatory filings, financial disclosures, or any use where an error can hurt somebody or breach a rule. Route those strings through human review before they ship, and keep the reviewer in your workflow rather than in your intentions. If your industry requires a certified or sworn translation, this Service does not produce one.

Do not put payment card numbers, bank credentials, government identifiers, passwords, biometric templates, or special category personal data (health, genetic, biometric identification, racial or ethnic origin, political opinions, religious beliefs, trade union membership, sex life or sexual orientation) into translatable content, screenshots or glossary entries. The Service is built for product copy, not for regulated records: it is not a PCI environment, not a HIPAA business associate, and not designed to hold that data. If you need to localize an interface that displays such data, submit the labels rather than the values.

Do not upload, store or transmit malware, ransomware, viruses, worms, logic bombs, cryptominers, or files crafted to exploit a parser. The import pipeline reads twenty file formats, including several XML dialects, and a malformed file aimed at a parser is an attack rather than a bug report. If you find a way to make the importer misbehave, that is a vulnerability report and the Vulnerability Disclosure policy tells you how to send it, safely and with a safe harbour.

Do not attempt to gain unauthorised access to the Service, to another organization, to another project, or to any account, system, network or data that is not yours. Do not probe, scan or test the vulnerability of the Service except as the Vulnerability Disclosure policy permits. Do not defeat, disable or circumvent authentication, rate limiting, quota enforcement, plan gating, token scoping or any other access control. Do not attempt to read, decompile, reverse engineer or extract our models, prompts, quality checks or system instructions, and do not use the Service to build or train a competing model.

Do not create multiple accounts or organizations to multiply a free allowance, to evade a cap, a suspension or a ban, or to obtain a trial more than once. Do not run automated traffic that exceeds published rate limits, that queues work solely to occupy capacity, or that destabilises the Service. Do not resell, sublicense, rent, timeshare or provide the Service as a service bureau to third parties, and do not expose it to your own end users as a general purpose translation API. Reselling is available under a written agreement; ask rather than assume. Where a plan describes seats, projects or locales as unlimited, that means we set no numeric cap rather than that any volume is acceptable: unlimited is for a real team doing real work, and we may contact you about usage that is wildly out of proportion to that, such as thousands of seats provisioned to accounts that never sign in, or projects created programmatically to shard an allowance. We will always talk to you before we act on that, and the word allowance is metered rather than unlimited on every plan, so this is about structure rather than volume.

Do not point a webhook, a connector or any other outbound feature at an address you do not control, at an internal or link-local address, or at a third party in order to send traffic they did not ask for. Do not use the Service as a relay, proxy, port scanner or amplifier. You are responsible for the security and the lawfulness of every endpoint you register, and for verifying the signature we send with each delivery.

Do not use the Service to compose, translate or deliver unsolicited bulk email, SMS or messaging content, and do not use it to localize a campaign that breaches applicable marketing law in the market it targets. Translating a lawful, consented campaign into more languages is fine. Using the Service to make an unlawful one bigger is not.

The open source grant is for genuinely open source work under a recognised licence. Do not use it to run commercial localization behind an open source shell, and do not split commercial work across granted organizations. The Founding Partner program is for honest recommendation. Do not self-refer, do not bid on our brand terms, do not place referral links in unsolicited email, comment spam, coupon farms or typosquatted domains, and do not misrepresent the product to earn a commission. The Founding Partner Terms carry the full list.

When you connect a repository, content platform, storage bucket or chat workspace, you are also bound by that provider’s terms, and you must have the right to grant us the access you grant. Do not use a connector to move content out of a system you are not permitted to export from, and do not use our storage connectors to host content that has nothing to do with localization.

If you believe someone is using Transglot in breach of this policy, write to hello@transglot.ai with the organization or project you are reporting, what you observed, and anything that helps us verify it. Copyright complaints go through the Copyright and DMCA policy, which has its own required form. Security findings go to security@transglot.ai. We do not disclose the identity of a reporter to the reported party unless the law compels us to.

Depending on what we find, we may contact you and ask you to fix it, remove or disable specific content, suspend a token, a project, a user or the whole organization, terminate the agreement, or report the matter to law enforcement. We aim to give notice and a chance to fix a problem first, and we make the response as narrow and as short as we reasonably can. We act without prior notice where the conduct is doing active harm, where it is unlawful, where notice would defeat the purpose, or where clause 04 applies. Suspension for a breach does not pause your billing.

If you think we got it wrong, reply to the notice or write to hello@transglot.ai within thirty days with the reasons. A person who was not involved in the original decision reviews it, and we will tell you the outcome. We restore access where an appeal succeeds. Clause 04 decisions are not appealable.

We do not read your strings for fun, and we do not run general surveillance of Customer Content. We do run automated abuse and integrity signals (rate limits, quota accounting, fraud checks, malformed input detection) and we look at specific content when a report, a legal obligation or a security incident requires it. What we log, and for how long, is described in the Privacy Policy.

We update this policy as the product and the abuse landscape change, and we change the effective date at the top when we do. A change that materially expands what is prohibited is announced under the notice provisions of the Terms of Service. Where a change is needed to comply with law or to address active abuse, it takes effect immediately.