Allow optional cookies for referral attribution, visit analytics, and Google Ads purchase measurement.
Read and confirm the safety rules before orderingPolicy version: 2026-10-09.v1 · 2026-10-09
Terms, Prohibited Uses and Safe-Use Commitment
Guishu provides LLM API aggregation and relay services for llm-token.cn and llm-agent.cc through cloud providers, Token factories and computing platforms. These rules apply to all models, languages, purchases, top-ups and downstream apps. Chinese law and other applicable mandatory law prevail.
01
Important: effects on your service and rights
Prohibited use may lead to rejected requests, restricted keys or accounts, suspension or termination, and legally required evidence preservation or reporting. Actions must be proportionate; balances, refunds and disputes follow applicable law and contract, without automatic forfeiture of lawful funds.
The platform audits content. Necessary evidence may include input text or context, with rule-based redaction, encryption and restricted review. We do not promise zero retention, full anonymization or screening of every input and output. Upstream providers have separate policies.
You are responsible under applicable law for your input sources, permissions, downstream use and unlawful conduct. Scrolling or checking a box does not waive platform duties or consumer rights, or validate an unenforceable disclaimer.
02
Prohibited content and conduct, including assistance or delegation
1. National security and secrets: threats to sovereignty, security, unity or national interests; unlawful subversion, separatism, espionage, or obtaining or disclosing state secrets and protected information.
2. Political impersonation: the platform does not offer customized impersonations, face swaps, voice clones, fabricated statements or misleading depictions of national leaders or political public figures, or fake official statements; deceptive, defamatory and manipulative uses are prohibited.
3. Heroes, martyrs and fabricated history: unlawful distortion or denigration of protected heroic deeds, names, images, reputation or honor; forged historical evidence or harmful misinformation impersonating historical political figures.
4. Terrorism and extremism: propaganda, recruitment, funding, planning or assistance for terrorist, extremist or other violent crimes.
5. Unauthorized intrusion: exploiting or accessing websites, apps, APIs, databases, cloud services, accounts or devices without authorization, including credential stuffing, account theft and takeover.
6. Malicious reverse engineering and cracking: analyzing apps, sites, protocols or clients to enable unauthorized attacks, data theft, license or security bypasses, cracking tools or attack payloads.
7. Malware and system damage: trojans, ransomware, botnets, destructive code, DDoS, exfiltration, traffic hijacking, data tampering or damage to computer information systems.
8. Unlawful proxies and VPNs: operating cross-border networking, proxy or VPN services without legally required permission, providing channels to evade regulation, blocks or safety auditing, or using them for unlawful acts.
9. Safeguard evasion: bypassing audits, suspensions, limits, authentication or payment; stealing or trafficking in others' keys, accounts or quota; knowingly supplying access for unlawful use.
11. Illegal finance and gambling: money laundering, unlawful fundraising, securities or financial activity, gambling promotion, or financial and technical assistance for unlawful transactions.
12. Drugs and controlled substances: manufacturing, refining, modifying, obtaining or trafficking in illicit drugs; unlawful synthesis, drug misuse or evasion of precursor-chemical controls.
13. Weapons, explosives and hazardous chemicals: actionable assistance to make, modify, acquire, use or conceal illegal weapons, explosive devices, toxic agents or other dangerous materials.
14. Nuclear and radiological safety: nuclear-weapons development, manufacture, acquisition or use; radioactive-material misuse or trafficking; sabotage of nuclear facilities or radiation safeguards.
15. Biosecurity: developing, modifying, acquiring or spreading harmful pathogens, toxins or biological weapons; compromising biosecurity or evading laboratory and quarantine controls.
16. Medical and health harm: unlicensed diagnosis, prescribing or human experiments; fabricated clinical or drug records; dangerous medication, self-harm or suicide instructions; unreviewed output controlling high-risk medical actions.
17. Pornography and sexual abuse: obscene sexual content, exploitation, sexual violence, pornography promotion, or non-consensual nudity and sexual deepfakes.
18. Harm to minors: any sexualization or exploitation of minors, including fictional or synthetic material; grooming, abuse, bullying, or inducing dangerous, unlawful or addictive conduct.
19. Privacy and personal safety: doxxing, stalking, harassment, threats or cyberbullying; processing sensitive face, voice, health, financial or location data without a lawful basis.
20. Hate and discrimination: inciting hatred, discrimination, persecution or violence against groups on grounds such as ethnicity, race, religion, gender or disability.
21. Intellectual property and trade secrets: infringing works, code, trademarks or databases, stealing secrets, bypassing rights protection or engaging in unfair competition.
22. Misinformation and manipulation: harmful rumors, fabricated news or emergency reports, manipulative political propaganda, inauthentic identity networks, fake engagement or reviews and deceptive marketing.
23. Critical systems and public safety: attacks or disruption against energy, communications, transport, finance, healthcare or industrial controls; assistance for physical harm, stalking or unlawful military uses.
24. AI labels and compliance evasion: maliciously removing, altering, forging or hiding AI labels; posing as a human or authoritative source; unlawful cross-border data disclosure or helping others evade these rules.
03
Use, transactions and downstream responsibilities
Platform restrictions are distinct from legal findings of wrongdoing. Restrictions on political impersonation or offensive automation do not mean that mentioning a person, lawful news, general political discussion, medical education, authorized interoperability research or lawful business networking is inherently illegal.
Security research requires clear authorization, defined assets and scope, and no harm to third parties. General scientific or medical information does not authorize high-risk experiments. Downstream apps, agents, resellers and teams must meet their own obligations.
You must have capacity to contract; use and spending by minors require legally necessary guardian consent. Submit only data you may lawfully process, protect keys and do not order under another person's identity.
Price, currency, quota, delivery and refunds follow the order page, enforceable terms and applicable law. Safety enforcement is not a blanket exclusion of refund, explanation, complaint or appeal rights.
When distributing AI content, make required disclosures and retain appropriate text, image, audio, video and metadata labels. A model response or an allowed request is not legal authorization.
04
Auditing, acceptance records and data handling
We keep necessary account, order, usage, security and audit records for billing, delivery, abuse prevention, disputes and legal duties. Model requests are not used for our own model training, advertising profiles or sale.
This order records the server confirmation time, policy version, document fingerprint, displayed language, site and confirmation method, linked to the order. Network information follows the Privacy Policy and existing security records; no additional identity document or unrelated data is required.
Redaction has limits and cannot remove every personal detail from free text; authorized review may decrypt evidence. Retention, deletion, legal disclosure and upstream handling are explained in the Privacy Policy.
05
Disclaimers and non-excludable responsibility
AI output can be wrong, incomplete, biased or outdated and cannot replace qualified professional judgment. Use human review proportionate to risk before publishing, running code or using output for medical, legal, financial or other high-risk decisions.
Unlawful users bear responsibility under applicable law. Platform liability depends on applicable law and actual fault; relay-only, research-only and checkbox language do not automatically remove it.
These terms do not exclude liability that cannot lawfully be excluded, including personal injury, property loss caused intentionally or through gross negligence, or mandatory consumer rights.
06
Explanations, reports and appeals
If you do not understand prohibited uses, restrictions, audit retention or liability, contact [email protected] for an explanation before agreeing. Do not send full keys, passwords or unrelated sensitive information.
You may decline and close this dialog; no new order is created without agreement. Each purchase or top-up requires active confirmation; a new policy or a pending order without a valid record requires confirmation again.
07
Legal and regulatory references
The following principal rules apply within their respective scope; not every rule applies directly to every activity. Links marked Catalog lead to the official current-law list; others lead to official legal texts or publications. Effective texts, territorial scope and competent authorities determine applicability.
Still have questions?
Reach our support team directly for anything about purchases, top-ups, delivery, or orders.