This is not legal advice. Plain-language terms, written to be honest and accurate.
GroundLab is a workspace where you reason with multiple AI models — to stress-test ideas, not to be told what to do. To use the platform you must be 18 years or older. The AI can be confidently wrong — verify before you act. Your sessions stay on your device. You pay for what you use, from a prepaid wallet, above and beyond the free tier, which is always open. The rules are below.
You must be 18 or over (we apply an age-assurance step at signup). GroundLab is an Australian product; access from sanctioned jurisdictions (e.g. those under Australian/US sanctions) is blocked. Attempting to bypass these access restrictions — using a VPN, proxy, or any other means to disguise your location — is a breach of these terms.
You commit that you will not use GroundLab to: generate prohibited content (e.g. child sexual abuse material, instructions for creating weapons or carrying out mass-casualty attacks, self-harm instructions); assert fabricated facts about real people as if true; harass, defame, or impersonate; infringe anyone's copyright or intellectual property — don't submit, reproduce, or publish material you don't hold the rights to; break the law or others' rights; or attempt to bypass the safety layer, our access restrictions, or our providers' policies. Reasoning about any topic, and through any persona, is the product — the line is the act, not the subject. For example: debating a topic is permitted; using the output to actually cause harm is not. The full Acceptable Use terms form part of these Terms (our Acceptable Use Policy).
Your session content lives on your device, not our servers. GroundLab acts as the conduit and keeps no copy. We hold information as set out in our Privacy Policy.
Inputs and AI outputs are screened to protect users and our access to the model providers. Enforcement is graduated: a notice, a withheld render/export, or — for the narrow catastrophic categories — a hard stop. Specialised debate (medical/legal/financial) is not blocked — it's covered by disclaimers, not suppression. Screening is disclosed openly; it is not a hidden filter.
What the screener gates — and what it doesn't: GroundLab's safety boundary is on the act, not the subject. Reasoning about difficult topics — violence, self-harm, contested politics, extremism — is the product. The screener blocks only a narrow set of acts: generating instructions for harm, producing illegal content, or material falling under catastrophic categories (child exploitation, instructions for creating weapons or carrying out mass-casualty attacks). The screener is powered by OpenAI's omni-moderation with a Google fallback, applied the same way for every user regardless of which AI debater is chosen.
Automated screening is imperfect — it can over-classify (block legitimate content) or under-classify (let something through). It is a safety layer, not a guarantee, and you remain ultimately responsible for what you submit and for how you use GroundLab's outputs.
Content here is AI-generated, and openly labelled as such. Recorded/published sessions carry an unremovable watermark and disclaimer (entertainment, not advice; AI can be confidently wrong — verify). Where a model requires attribution (e.g. "Built with Llama"), it is shown. GroundLab is not designed to produce content that mimics real news or impersonates real people, and such use is prohibited (§3).
The outputs you create (transcripts, rendered episodes) are yours to use and commercialise, to the extent the law allows. Note that under current Australian law, copyright generally requires human authorship, so purely AI-generated material may not attract copyright protection — if you intend to rely on or commercialise outputs, consider taking your own legal advice. You're responsible for what you publish, and GroundLab isn't liable for content you distribute.
GroundLab is provided "as is", without warranties, to the maximum extent permitted by law. Model availability can change — a provider may retire a model; we'll substitute a model of comparable capability where possible, and notify you of material changes in advance. Nothing in these terms excludes rights you have under the Australian Consumer Law that cannot be excluded. Beyond that, our liability is limited to the maximum extent the law allows.
Your responsibility. If you breach these terms or misuse GroundLab and that directly causes a third-party claim or loss to GroundLab, you may be responsible for the reasonable resulting costs, to the extent permitted by the Australian Consumer Law — this is deliberately narrow (an over-broad indemnity against a consumer can itself be unenforceable), not a blanket clause.
We may suspend or close an account for AUP violations, repeated safety trips, non-payment, or fraud, with reasonable notice where practicable, and will outline any billing effects (§8). You may stop and close your account at any time; account deletion removes your account, billing records, and personal data (Privacy Policy §7) — but never your session content, which was always on your device (§4) — except for (a) moderation evidence (retained per Privacy Policy §5) and (b) feedback you submitted, which is retained but anonymised (your contact details are removed).
Tell us what's working or not — and raise concerns — via our Feedback & Complaints form. A privacy concern can also go to the Office of the Australian Information Commissioner (Privacy Policy §12).
We may update these terms; material changes are surfaced at sign-in (the versioned-acceptance rail), not buried, and may require your re-acceptance. The "last updated" date below reflects the current version.
These terms are governed by the law of Queensland, Australia. GroundLab (groundlab.com.au) is operated by sole trader ABN 88 609 151 945, registered in Qld, Australia. Contact: our feedback form.
Last updated: 2026-06-20 · v1.0.