A creator's old overseas content can become tomorrow's China-market crisis—but “彻底审查” must still be lawful, relevant, accurate, and fair.
Before investing in localization, commercial introductions, account growth, or exclusivity, an agency should understand the creator's authentic public history. The purpose is to identify concrete signing and market-entry risk—not to infer private beliefs, demand political conformity, or punish lawful personal expression unrelated to the role.
Sound screening
Verified identity × complete declared-account map × relevant evidence × cultural context × human judgment × privacy controls
Section 1
Define the Decision, Scope, and Rules Before Searching
Document first
- Role, markets, platforms and public-facing duties
- Specific legal, brand, platform and cultural risks
- Accounts and time period reasonably necessary
- Sources permitted and sources prohibited
- Severity definitions and decision owners
- Retention, access, deletion and appeal process
Never use as a shortcut
- Nationality, ethnicity, religion or other protected traits
- Private political beliefs or inferred ideology
- Family, health or intimate-life speculation
- Anonymous accusations without corroboration
- Guilt by association or follower behavior
- A brand's discomfort without a role-related rationale
Section 2
Build a Verified Overseas Account Inventory
Self-disclosure
Ask the creator to list current and former public usernames, aliases, channels, podcasts, blogs, storefronts, community profiles, and materially controlled accounts.
Identity resolution
Match through linked bios, official websites, cross-posts, historical usernames, consistent content, and creator confirmation—never name similarity alone.
Ownership labeling
Separate creator-owned, employer-owned, fan, parody, impersonation, hacked, shared, dormant, and deleted accounts.
Coverage record
Log platform, handle, URL, owner confidence, active dates, language, geography, review date, and unresolved gaps.
Section 3
Review Public History with a Repeatable Protocol
Review themes
- China/Taiwan/Hong Kong/Tibet and territorial references
- National symbols, maps, sovereignty and geopolitical claims
- Racism, hate, harassment, violence or extremist support
- Religion, ethnicity, gender and cultural stereotypes
- Fraud, undisclosed advertising, fake engagement or impersonation
- Illegal/dangerous conduct and harmful misinformation
- Competitor, exclusivity and conflict disclosures
- Account strikes, bans, hacks and coordinated controversy
Evidence protocol
- Open the original post—not only a screenshot
- Capture URL, platform, handle and timestamp
- Preserve surrounding thread/video and translation
- Distinguish authoring, quoting, reporting and criticism
- Record edit/deletion and archive limits
- Corroborate serious allegations independently
- Label fact, interpretation and unknown separately
- Use a second reviewer for elevated findings
Finding record
Who + what + when + original context + reliable translation + authenticity confidence + role relevance + applicable rule + creator response
Section 4
Classify Conduct—Not People
| Level | Illustrative evidence | Response |
|---|---|---|
| Critical | Credible unlawful conduct, sanctions issue, violent/extremist support, coordinated deception, or content that counsel confirms creates an unacceptable China-market exposure | Pause; preserve evidence; legal/senior review; do not sign until resolved |
| High | Repeated hateful or discriminatory attacks, material fraud, undisclosed serious conflicts, dangerous misinformation, or sustained conduct incompatible with the role | Second reviewer; creator response; documented accept/mitigate/decline decision |
| Medium | Isolated offensive material, ambiguous geopolitical/cultural content, recurring poor judgment, undisclosed legacy account, or meaningful brand mismatch | Context interview; remediation plan; contractual/content controls where appropriate |
| Low | Old minor controversy, ordinary criticism, satire clearly understood in context, or taste mismatch without material role-related risk | Record minimally; avoid overreaction; proceed if otherwise qualified |
Severity
Potential harm
Legal / human / brand
Confidence
Source quality
Identity / authenticity
Recency
When + pattern
Not age alone
Relevance
Role / market
Exposure and control
Section 5
Preserve Language, Time, and Cultural Context
Questions for every finding
- Is the account genuinely controlled by this creator?
- Is the content authentic and complete?
- Was it authored, quoted, criticized, satirical or newsworthy?
- What did the words/symbols mean in that language and time?
- Was there correction, apology, learning or repeated conduct?
- What actual duty, audience or market risk does it affect?
Translation controls
- Retain original language beside translation
- Use qualified native/cultural review for serious cases
- Explain slang, reclaimed terms, irony and historic context
- Do not rely on machine translation for final decisions
- Separate literal meaning from likely audience interpretation
- Record uncertainty instead of forcing certainty
Section 6
Use Notice, Response, and Human Decision Gates
Verify
A trained reviewer confirms identity, source, context, translation, relevance, and severity.
Notify
Show the creator the material concern in enough detail to understand and respond; protect third parties and confidential information.
Hear
Record explanation, contrary evidence, account compromise, impersonation, remediation, and any factual dispute.
Escalate
Legal/compliance and a culturally qualified reviewer examine high or critical findings; conflicts are disclosed.
Decide
Choose sign, sign with proportionate controls, defer pending evidence, or decline—with a documented, role-related rationale.
Appeal
Provide a route to correct identity, translation, authenticity, or material factual errors.
Section 7
Treat Public Content as Personal Information
Data controls
- Provide a clear screening notice
- Identify the processing basis with counsel
- Collect only necessary public evidence
- Apply stricter controls to sensitive information
- Restrict access to named reviewers
- Encrypt exports and log access
- Set short retention/deletion rules
- Support correction and applicable rights
Vendor controls
- Define sources and prohibit covert/private collection
- Ban data resale and unrelated model training
- Test identity matching and translation accuracy
- Require security and breach notification
- Control subprocessors and cross-border transfers
- Audit false positives and disparate outcomes
- Require human review and explainability
- Ensure deletion and return on termination
Section 8
Monitor Proportionately After Signing
Event-based refresh
- New market, platform, category or major brand
- Creator discloses another account or identity
- Credible controversy or platform enforcement
- Account hack, impersonation or ownership change
- Material legal/regulatory change
- Renewal for a high-exposure relationship
Respectful boundaries
- Explain monitoring in policy/contract
- Focus on public role-related conduct
- Avoid continuous surveillance by default
- Do not demand personal-account credentials
- Re-verify source before escalation
- Use the same response and appeal process
Monitoring rule
Frequency and depth ∝ current exposure, credible trigger, and potential harm—not curiosity or control
Section 9
Pre-Signing Review Checklist
Scope & collection
- Role and risk criteria approved
- Creator notice delivered
- Declared accounts collected
- Identity and ownership verified
- Only permitted public sources used
- Coverage gaps documented
Evidence & judgment
- Original context and URLs retained
- Material translation independently checked
- Rumor and fact separated
- Severity/confidence/relevance scored
- Elevated findings second-reviewed
- Creator response considered
Decision & contract
- Rationale is role-related and consistent
- Legal/cultural escalation completed
- Mitigations are specific and proportionate
- No forced concealment or vague morality clause
- Appeal/correction route provided
- Approval owner signed the record
Privacy & lifecycle
- Access is need-to-know
- Sensitive data minimized
- Vendor and transfer controls checked
- Retention/deletion date set
- Monitoring triggers disclosed
- Audit and incident process ready
Section 10
Official China Guardrails
This operational guide is not legal advice. The right scope depends on the creator's location and status, the agency's role, markets, data flows, and the specific decision.
Personal Information Protection Law of the PRC ↗
Official NPC text on lawful processing, notice, sensitive information, automated decisions, public personal information, rights, and cross-border handling.
Provisions on the Governance of the Online Information Content Ecosystem ↗
Official CAC rules addressing unlawful and harmful online content and the responsibilities of content producers, platforms, and users.
Provisions on the Administration of Algorithmic Recommendations ↗
Official CAC rules covering algorithm governance, user labels, personal-information protection, human intervention, transparency, and complaints.
Search broadly enough to find real risk. Decide narrowly enough to remain fair.
A defensible background check maps verified public accounts, preserves context, distinguishes evidence from inference, gives the creator a voice, and limits data and decisions to what the role genuinely requires.
“Review content, not identity. Verify context, not rumor. Let trained humans—not keywords—own the decision.”