Safety & Review Prompts

Public Content & Brand Safety Review

Review public-facing copy before publication for harmful ambiguity, unsupported claims, privacy leaks, disclosure gaps, audience risk, and brand damage.

FreeClaudeChatGPT
Best for

Founders, marketers, creators, community managers, and small teams reviewing landing pages, announcements, social posts, ads, emails, and public statements.

Suitable LLM groups
FrontierReasoning
Download Prompt.md
Prompt
Act as a careful public-content and brand-safety reviewer. Review the draft below before it is posted, emailed, advertised, published, or sent to customers. Preserve the intended message while identifying preventable risk.

Draft content:
{{draft_content}}

Brand, audience, channel, and goal:
{{brand_audience_channel_goal}}

Required claims, sources, disclosures, and constraints:
{{claims_sources_disclosures_constraints}}

Review it using this structure:

1. Purpose and audience fit: summarize what the content asks the audience to believe, feel, or do.
2. Claim review: flag unsupported facts, guarantees, superlatives, comparisons, performance claims, statistics, testimonials, or implied promises. Mark claims that need evidence or qualified review.
3. Safety and harm review: identify language that could encourage unsafe behavior, target vulnerable people, normalize harassment, disclose private information, or create foreseeable misuse.
4. Brand risk: flag misleading urgency, fake scarcity, manipulative framing, insensitive language, accidental insults, tone mismatch, reputational spillover, or claims that conflict with stated values.
5. Disclosure and attribution: identify missing sponsorship, affiliate, AI-assistance, source, permission, testimonial, limitation, or eligibility disclosures where context suggests they may be needed. Do not invent legal requirements.
6. Privacy and confidentiality: flag names, screenshots, customer details, internal metrics, private messages, location clues, or metadata that should not be public.
7. Severity-ranked changes: label each issue Block publication, Fix before publication, or Optional improvement.
8. Safer rewrite: produce a revised version that preserves the goal, removes unsupported certainty, and adds only disclosures supported by the provided context.

Rules: Do not make legal compliance claims. Do not invent evidence, permissions, testimonials, or disclosures. For regulated products, health claims, financial claims, political advertising, children, or high-risk audiences, recommend qualified review before publication.

How to use

  1. Paste the draft and state the audience, channel, objective, evidence, permissions, and required disclosures.
  2. Resolve every block-publication item before release.
  3. Verify factual claims and obtain permission for names, quotes, screenshots, and customer stories.
  4. Use specialist review for regulated, political, health, financial, or child-directed content.

Example input

Draft: 'Our AI hiring tool eliminates bias and guarantees the best candidate every time. Sign up today before the price doubles forever.' Channel: LinkedIn ad. Evidence: internal demo only; no independent evaluation.

Example output

Block publication. The claims 'eliminates bias,' 'guarantees,' and 'best candidate every time' are absolute and unsupported. The permanent price-doubling statement may be misleading if no documented pricing decision exists. Safer rewrite: 'Use structured AI assistance to organize applicant information for human review. Join the early-access list to learn about current pricing and evaluation results.' Add clear information about human oversight and avoid fairness claims until supported by appropriate testing.

Customization tips

  • Add your brand voice and prohibited phrases to make the rewrite more consistent.
  • Ask for separate reviews of claims, tone, privacy, and disclosure when the content is complex.
  • Provide platform limits and ad-policy constraints when reviewing paid campaigns.

Tags

#brand safety#content review#claim review#privacy#publication checklist

FAQ

Does this prompt replace legal review of advertising claims?
No. It flags common risk patterns and evidence gaps but cannot determine legal compliance for a specific jurisdiction or product.
Can it review screenshots and visual assets?
It can review a detailed description or provided visual context, but rights, likeness, and trademark questions may need specialist review.
What counts as an unsupported claim?
A factual or performance statement is unsupported when the provided evidence does not adequately establish it for the audience and context.
Should all AI-assisted content include an AI disclosure?
Not necessarily. Disclosure needs depend on context, platform rules, audience expectations, and applicable requirements; the prompt should flag the question without inventing a rule.
How are publication blockers different from optional edits?
Blockers create material safety, privacy, deception, or reputational risk, while optional edits mainly improve clarity, tone, or polish.
Free

Privacy-Safe Data Sharing Review

Review text, files, screenshots, or datasets before sharing them with an AI tool, vendor, colleague, or public audience.

ClaudeChatGPT
#data privacy#redaction#data minimization
Free

AI Output Fact-Check & Evidence Audit

Audit an AI-generated answer for unsupported claims, weak citations, invented details, outdated facts, and misleading certainty before reuse.

ClaudeChatGPT
#fact checking#hallucination review#citation audit
Free

High-Risk Marketing Claims Review

Review marketing copy for guarantees, health or financial implications, fake urgency, weak evidence, hidden conditions, and claims that need specialist approval.

ClaudeChatGPT
#marketing claims#advertising review#consumer safety

Related Skills

Safety, Privacy & ComplianceFree

Sponsored Content Disclosure and Brand Safety Reviewer

Review sponsored creator content for clear disclosure, claim risk, audience trust, privacy, brand fit, prohibited language, and approval needs.

ClaudeChatGPT
#sponsored content#disclosure review#brand safety
Testing & Quality ChecksFree

Social Content Preflight QA Reviewer

Review social content before publishing for accuracy, claims, links, tags, captions, subtitles, visuals, disclosure, accessibility, and platform fit.

ClaudeChatGPT
#social media QA#content review#publishing checklist
Safety, Privacy & ComplianceFree

AI Governance & Compliance Checklist

Reviews an AI-related workflow, tool, or output against a practical governance checklist covering data privacy, bias, accountability, and transparency.

ClaudeChatGPTCursor
#AI governance#compliance#responsible AI

Related Articles

Article · AI Safety Privacy and Risk

AI-Generated Content Needs Provenance and Human Grounding

AI safety is becoming more practical, focused on permissions, monitoring, refusal quality, provenance, and human review.

Jul 9, 20267 min read
Read AI-Generated Content Needs Provenance and Human Grounding
Article · AI Skills

Are AI skills safe to download and use?

AI skills can be useful, but they should be treated like operational instructions. Users should review permissions, external actions, scripts, and unclear triggers before trusting them.

Jul 9, 20268 min read
Read Are AI skills safe to download and use?
Article · AI Safety

How do I redact sensitive information before using ChatGPT or Claude?

Redaction means removing or masking personal, financial, customer, credential, health, and confidential business details before sharing text with AI.

Jul 9, 20268 min read
Read How do I redact sensitive information before using ChatGPT or Claude?