Safe conversational advertising
Why LLM ad safety matters
LLM ads are clearly labeled as ads. But labeling alone is not enough. In conversational interfaces, the boundary between content and persuasion becomes thinner. Users often perceive AI responses as guidance from a trusted assistant, not a traditional advertisement. That means even ethically placed ads must be designed with care, limits, and user first principles.
We believe LLM advertising should enhance decision making not manipulate it.
Advertiser Elegibility
Every brand that enters the Thrad marketplace is reviewed for business legitimacy, legal compliance, and creative standards before campaigns are activated.
Publisher Trust
Every LLM app or platform using Thrad must provide users with ad-visibility controls and abide by clear disclosure requirements.
Consumers
Respect and transparency
Publishers
Quality monetization and protection
Advertisers
Safe and ethical advertisements
Our commitment to consumers, publishers, and advertisers:
Our Commitment to Advertisers
Your brand appears only in appropriate contexts. We exclude conversations involving mental health crises, medical distress, financial vulnerability, and other sensitive situations. Every publisher is vetted for legitimacy and safety standards. Your campaign data is encrypted and never shared. In return, we require all advertisers to meet the same quality and accountability standards.
Our Commitment to Publishers & Their Users
We vet every advertiser for legitimacy and compliance before campaigns launch. Sensitive conversations are automatically protected—no advertiser can override these exclusions. All ads carry mandatory disclosure labels. We process only pseudonymized, non-sensitive conversational signals for relevance. No health data, biometrics, political affiliation, cookies, or device fingerprinting.
Our Commitment to End Users
Ads are always distinguishable from AI responses. We prohibit manipulative tactics, false urgency, and emotional pressure. Users in vulnerable moments will not see advertising—this is enforced automatically. Conversational data is not used for targeting; ad selection is contextual only.
Publisher Context Standards
Eligible Conversation Contexts
Advertising may be served in conversations that are informational, research-oriented, task-focused, or commercial in nature, where the user is actively seeking product or service information.
Ineligible Conversation Contexts
Advertising must not be served in conversations involving:
Mental health, self-harm, suicidal ideation, or emotional crisis
Medical diagnosis, treatment decisions, or emergency health situations
Legal proceedings, criminal matters, or situations involving potential legal jeopardy
Financial distress, debt crisis, or situations of economic vulnerability
Bereavement, trauma, or acute personal distress
Religious, spiritual, or deeply personal belief discussions
Political opinion, electoral matters, or civic participation•
Content involving minors in any sensitive context
Contextual processing uses pseudoanonymised conversation signals only. No sensitive personal data categories as defined under GDPR Article 9 are used in ad selection or targeting.
Publisher Eligibility
Publisher platforms must represent legitimate, operational LLM-powered applications with genuine user bases and must accurately represent their platform type, user demographics, and inventory characteristics to Thrad. Publishers must operate in compliance with applicable data protection laws and maintain user-facing privacy disclosures consistent with Thrad's data processing role. Publishers must not inject Thrad ad placements into third-party content or misrepresent the context in which ads appear. All impressions must originate from genuine human engagement; automated traffic and incentivised ad views are prohibited.
Prohibited Advertising Categories
The following categories are incompatible with advertising through Thrad under any circumstances:
Sexually explicit or obscene material.
Discrimination, hate speech, or harassment.
Extreme violence, gore, or glorification of harm.
False, misleading, or disinformation content.
Unlicensed gambling, speculative schemes, or predatory financial products.
Counterfeit or illegally sold goods.
Controlled substances, drug paraphernalia, or illegal drug facilitation.
Malware, hacking tools, or exploitation services.
Piracy-enabling platforms or IP theft tools.
Non-consensual synthetic media, impersonation, or source misattribution.
Incentivized engagement, click farms, or manufactured traffic.
Content exploiting, sexualizing, or targeting minors.
Categories Requiring Prior Approval
The following categories may be permitted with explicit prior approval from Thrad's policy team.
Approval is per-campaign and does not extend to the category broadly.
Pharmaceutical and health products, including supplements and medical devices
Alcoholic beverages
Tobacco, nicotine, and vaping products
Financial products and services, including digital assets and investment platforms
Political campaigns and issue advocacy
Weapons and related accessories (where permitted by law)
Matchmaking, dating, and relationship platforms
Content directed at or likely to reach audiences under 18
To request approval for a restricted category, contact contact@thrad.ai before campaign submission.
Enforcement
Thrad uses both automated monitoring and human review to maintain marketplace quality. Violations may result in:
Written notification with required remediation steps
Temporary campaign pause or inventory suspension pending review
Permanent removal from the Thrad platform
Revenue adjustment or withholding for non-compliant traffic
Legal remedies where appropriate
Severe or repeated violations may result in immediate action without prior warning. Enforcement review requests can be submitted to contact@thrad.ai. Thrad commits to acknowledging requests within two business days.We believe LLM advertising should enhance decision making not manipulate it.
Conversational AI Advertising Standards
Clear Differentiation
Sponsored content must be visually or contextually distinguishable from the host AI's generated responses. Ads should feel native to the interface but must never be mistakable for organic AI output.
Mandatory Labelling
All advertising must carry a clear, prominent disclosure that it is paid advertising. Thrad requires publisher partners to implement its standard disclosure label. Campaigns that suppress or obscure this labelling will not be permitted.
No Manufactured Urgency or Manipulation
Using the conversational AI context to manufacture urgency, simulate emotional pressure, or manipulate user behavior toward commercial actions is prohibited. This includes:
Copy that mimics distress or urgency signals from the AI itself
Targeting based on detected emotional states or vulnerability indicators
Offers framed as exclusive or time-limited without genuine basis
Exploiting a user's apparent mood, state, or circumstances for commercial gain
Contextual Relevance
Advertisements should be relevant to the context of the user's conversation. Thrad reserves discretion to decline or pause placements that would appear incongruous, disruptive, or inappropriate, regardless of whether they meet other category standards.
Responsible AI Association
Advertising inventory must not be associated with AI systems that generate harmful content, bypass safety mechanisms, engage in deceptive practices, or violate applicable AI governance frameworks.
Data & Privacy Commitments
What Thrad Processes
For ad selection and performance measurement, Thrad processes aggregated performance metrics and pseudoanonymised conversation excerpts used to evaluate contextual ad relevance. Direct identifiers are removed prior to processing.
Thrad does not collect, process, or use:
Sensitive personal data categories under GDPR Article 9, including health data, biometric data, financial account information, or political or religious affiliation
Directly identifying end-user information such as names, email addresses, or persistent device identifiers
Cookies, device fingerprinting, or tracking pixels
Profiling
Conversation data is not used for profiling or behavioral targeting by default. Profiling may only occur where explicitly agreed with the publisher in advance and in writing, limited to pseudoanonymised segments with no legal or similarly significant effects on individuals.
Legal Basis & Roles
Thrad processes end-user data under legitimate interest (Article 6(1)(f) GDPR), strictly limited to improving ad relevance and effectiveness. Publisher partners act as data controllers; Thrad acts as a data processor on their behalf. For U.S. purposes, Thrad acts as a Service Provider under CCPA/CPRA.
Infrastructure & Security
All data is stored on AWS with encryption at rest (AWS KMS) and in transit (TLS/HTTPS). Hosting is US-based. EU/UK data transfers are governed by Standard Contractual Clauses.
Retention
Advertiser and campaign data is retained for up to 12 months following account inactivity or contract termination. Pseudoanonymised conversation data is retained only as long as needed for reporting, then deleted. Deletion requests: contact@thrad.ai.
Contact
For questions about these Standards, advertiser eligibility, restricted category approvals, or enforcement matters, contact Thrad at contact@thrad.ai or visit thrad.ai. These Standards are reviewed annually or following material changes to Thrad's platform, regulatory environment, or business operations.