AI Transparency & Disclosure

1. You are interacting with an AI system

• Bee is an artificial-intelligence system. Responses in the Bee Workspace, through the Bee API, in editor integrations, and via the hosted MCP server are machine-generated, not written by a human.
• If you build a product on the Bee API and expose AI-generated content or an AI interlocutor to your own end users, you are responsible for making an equivalent disclosure to them where the law of their jurisdiction requires it.

2. Capabilities and limitations

• Bee is a large-language-model platform. Like all LLM systems, it can produce output that is incorrect, incomplete, outdated, or misleading (including plausible-sounding fabrications), and its behaviour varies with the input it is given.
• Do not rely on Bee output as the sole basis for consequential decisions — medical, legal, financial, safety-critical, or decisions producing legal or similarly significant effects on a person. Keep a qualified human in the loop. Restricted domains (Tier-3) additionally require explicit in-product acknowledgement under the Terms.
• The capabilities each tier actually serves are published at bee.heossi.com/models and derive from a live-verified capability register — the published matrix reflects what is proven against the running production backend, not intent.

3. AI-generated media (synthetic content)

• Media created with Bee's generation features (for example image generation in the workspace) is synthetic — it is AI-generated content.
• Embedded machine-readable provenance marking for generated media (Content-Credentials-style signed metadata) is in development and is not yet embedded in generated files. Until it ships, you must not present Bee-generated media as human-created, or as an authentic depiction of a real person, place, or event.
• Using Bee to create deceptive deepfakes, to impersonate a real person without consent, or to remove or falsify provenance information is prohibited under the Acceptable Use Policy.

4. Machine-verifiable product facts

Bee publishes its product facts (tiers, capabilities, pricing, recent releases) in a machine-readable contract at bee.heossi.com/facts.json, dual-signed with ML-DSA-65 (NIST FIPS 204, post-quantum) and Ed25519. The verification keys are published at bee.heossi.com/trust/facts/keys.json. Anyone — a customer, a regulator, or an autonomous agent — can verify that a claimed Bee fact is authentic and untampered instead of trusting a screenshot.

5. Practices we do not offer and do not permit

• We do not offer, and you may not use Bee for: social scoring of natural persons; exploitation of vulnerabilities of specific groups (age, disability, social or economic situation); purposefully manipulative or deceptive techniques that materially distort a person's behaviour; inference of emotions of natural persons in workplaces or educational institutions (except for medical or safety reasons); biometric categorisation to deduce sensitive attributes; or untargeted scraping of facial images to build recognition databases.
• These prohibitions align with the practices prohibited by the EU AI Act and are incorporated into the Acceptable Use Policy, which controls in case of conflict.

6. Human oversight

Bee is designed to operate under human oversight: outputs are delivered to a human (or to a system a human configures), restricted domains require explicit acknowledgement, and safety filters run in front of every chat completion. For high-impact deployments, we recommend documented human-review checkpoints and encourage customers to contact us for deployment guidance.

7. Provider identity and regulatory contact

Bee is provided by HEOSSI (Pte.) Ltd. (UEN 202532790K), a private limited company incorporated in Singapore. Regulatory and legal enquiries: bee-legal@heossi.com. Serious incidents involving the service's security: bee-security@heossi.com. Data-protection matters: bee-privacy@heossi.com.

8. Updates

Material changes to this statement are announced via the changelog and (for account holders) by email before they take effect, consistent with the Terms.