Effective date: 29 July 2026
1. Purpose
This notice applies where Enterprise Corporated provides or displays artificial-intelligence, machine-learning, generative, classification, agentic, search, recommendation, automation or decision-support functions.
2. Nature of AI output
AI output is probabilistic and may be inaccurate, incomplete, outdated, non-unique, inconsistent or affected by data and model limitations. Output may contain errors, invented citations, insecure code, biased assumptions, inappropriate recommendations or material that resembles third-party content.
AI output is a candidate or draft unless an authorised human has reviewed and approved it for the stated purpose. The user must apply independent judgement and appropriate professional review.
3. No professional substitution
AI functions do not replace a lawyer, patent attorney, engineer, cybersecurity assessor, financial adviser, medical practitioner, forensic expert or other qualified professional. They do not provide a patentability, validity, infringement, freedom-to-operate, regulatory, security or investment opinion.
4. Human authority and significant decisions
Enterprise Corporated's intended governance model preserves human authority over material release, publication, licensing, legal use and significant decisions. We do not intend to rely solely on automated processing for decisions that produce legal or similarly significant effects without appropriate notice, lawful basis, safeguards and human review.
Users must not delegate final responsibility for high-impact decisions to an AI function unless an expressly approved service and governance process applies.
5. Input restrictions
Do not enter:
- private keys, passwords, seed phrases, API tokens or secret credentials;
- classified, export-controlled or unlawfully obtained material;
- privileged legal communications unless an approved confidential service expressly supports them;
- sensitive personal information unless necessary, lawful and authorised;
- unpublished enabling patent or trade-secret material through a public or unapproved model; or
- content that infringes rights or violates law.
6. Customer ownership and processing
Customers retain rights they lawfully hold in inputs. Rights in outputs depend on applicable law, third-party model terms, originality and the signed agreement. Enterprise Corporated does not guarantee that an AI output is protectable, exclusive or non-infringing.
Customer Confidential Information will not intentionally be used to train a general-purpose model for unrelated customers unless expressly authorised in writing. Before confidential use, the actual model provider, retention, training setting, region, subprocessors and deletion controls must be verified.
7. Third-party AI providers
A service may route requests to an external AI provider. The service-specific notice must identify the provider category, whether content is retained or used for provider improvement, applicable region, and available controls. Third-party terms may apply.
8. Safety and prohibited use
AI functions must not be used to facilitate unlawful access, malicious code deployment, fraud, impersonation, harassment, exploitation, discrimination, child abuse, weapons harm, self-harm encouragement, privacy invasion, rights infringement, deceptive evidence, false professional credentials or prohibited high-risk activity.
Defensive cybersecurity, forensic and research uses require lawful authority, controlled scope and compliance with the Acceptable Use Policy.
9. Intellectual property and provenance
AI assistance may affect authorship, inventorship, copyright and patent analysis. Users must accurately record material human contributions, model assistance, source material, review and approval where those facts are legally or evidentially relevant.
Hashes, signatures, timestamps and AI interaction records can support provenance but do not conclusively determine legal inventorship, authorship or ownership.
10. Testing, monitoring and change
AI systems may be tested, monitored, logged, filtered, rate-limited, evaluated and updated for quality, safety, security and compliance. Model behaviour can change when a provider or configuration changes. Material deployments should use version, test, approval and rollback controls.
11. Complaints and review
Report an AI output concern, rights issue or request for human review to [email protected]. Provide the date, service, relevant output and reason, excluding private keys or unnecessary sensitive information.