Acceptable Use Policy

Defines prohibited and restricted uses of WEBSPACEAI Services.

Applies to all users across all products and API access.

• Prohibited Use / High-Risk Use — as previously defined; cross-referenced in AI Safety Framework Risk Levels Doc. 24.
• CBRN-Adjacent Query — any request seeking synthesis routes, weaponization guidance, acquisition pathways, or uplift for chemical, biological, radiological, or nuclear weapons, including dual-use research framed as benign.

Users may not use the Services to generate content that facilitates violence, terrorism, or weapons development; exploit or endanger minors; conduct fraud, phishing, or impersonation; violate IP or privacy rights; generate non-consensual intimate imagery or defamatory synthetic media; circumvent safety mitigations; or operate autonomous agents evading human oversight.

⟳Explicit CBRN Red Lines The following are explicitly and categorically prohibited with no exceptions, including for research or educational framing: (a) synthesis or acquisition instructions for biological, chemical, radiological, or nuclear weapons or their precursors; (b) uplift to pathogen enhancement, gain-of-function guidance outside licensed institutional contexts, or novel toxin design; (c) technical uplift to cyberweapons targeting critical infrastructure. Dual-use queries (e.g., legitimate virology, pharmacology, nuclear engineering education) are routed through an expert review gate: outputs in these domains are checked against a maintained restricted-topic classifier, and borderline cases are escalated to the AI Safety Research Lead's domain-expert panel before a response is generated or the pattern is permanently blocked.

⟳Foreseeable Use Exclusion, per EU AI Act guidance Each product's documentation must expressly and consistently exclude out-of-scope high-risk uses (e.g., medical diagnosis, credit scoring, employment decisions, law enforcement) unless a specific enterprise agreement and compliance review authorizes such use, to avoid default high-risk classification under emerging AI regulation.

High-risk uses (medical, legal, financial, critical-infrastructure) require human review and enterprise agreements per the AI Usage Policy.

Trust & Safety owns enforcement; Product implements guardrails and classifiers; Legal reviews edge cases; AI Safety Research Lead chairs the CBRN expert review panel.

Automated and human review monitor flagged content. CBRN classifier accuracy is audited quarterly.

Content removal, rate limiting, suspension, termination, or law enforcement referral. CBRN violations trigger immediate account termination and, where required by law, mandatory reporting.

abuse@webspaceai.in | trust@webspaceai.in

No exceptions permitted for CBRN red lines. Other narrowly scoped exceptions require Trust & Safety approval within 10 business days, documented in the Exceptions Register.

Reviewed quarterly (increased from semi-annual) given evolving threat landscape.

trust@webspaceai.in | abuse@webspaceai.in | safety@webspaceai.in