MTA
Defines MTA principles for ethical AI deployment in SMM operations: transparency, human oversight, bias mitigation, DPDP compliance, deepfake prohibition, and C2PA Content Credentials. Forms part of every SMM retainer, proposal, and client SOW.
Last updated: April 2026
This policy ensures ethical, transparent, and trustworthy social media content for clients (gyms, healthcare providers, local businesses in Rajasthan). AI must serve the client goals while protecting audience trust, complying with law, and avoiding bias or harm. Applies to all SMM services: content creation, Reels/Shorts, carousels, captions, images, videos, ad creatives, community replies, and AI-automated workflows.
All AI-generated or AI-assisted content must be clearly labeled as Synthetically Generated Information (SGI) per IT Amendment Rules 2026. Implementation: Prominent caption line "AI-assisted content • Edited by MTA" plus platform-native labels (Instagram "AI-generated", YouTube disclosure) where available. No exceptions for "minor edits".
100% of AI output receives human review and final approval before posting. You or designated team member reviews every post. Audit log maintained (prompt + output + editor name) for compliance verification.
Content must not discriminate by gender, age, caste, religion, location (urban/rural Rajasthan), or language. Prompts tested on diverse audience segments. Weekly manual bias check on gym/healthcare campaigns to ensure fair representation.
Never feed client personal data (gym member lists, patient contacts, app users) into AI tools without explicit written consent. Anonymize data before n8n/AI workflow nodes. Data minimization: only necessary information used for processing. Full DPDP Act 2023 compliance required.
All health/fitness/medical claims must be fact-checked against reliable sources. No deepfakes or misleading "before-after" transformations. Healthcare content requires doctor/client sign-off before publication. Claims limited to verified features only.
Deepfakes (SGI that appears real and indistinguishable) are prohibited for client testimonials or transformations unless: (1) client provides explicit verifiable consent, (2) final video is human-edited, and (3) content is clearly labeled. Deepfake detection (Resemble AI + Hive Moderation) required on all media before posting.
All AI-assisted visual content must embed C2PA v2.4 Content Credentials before publication. This includes cryptographic manifest with creator claim, AI assertions, edit history, and content bindings. Supported formats: JPEG, PNG, WebP, AVIF, HEIC, MP4, MOV. Tools with C2PA support: Canva Magic Studio, Adobe, Microsoft.
Buffer AI / Predis.ai / Canva Magic: Always add manual disclosure text; human review of performance prediction scores. Claude / Gemini: Prompts must include "Rajasthan audience, Hindi/English, transparent tone". n8n workflows: Anonymization step required before any AI node.
MTA remains responsible for all AI-assisted output. Grievance mechanism: clients contact manglamtechnicalagency@gmail.com within 24 hours for response. Internal reviewer assigned per campaign. Audit logs kept for 2 years per DPDP requirements.
Any suspected violation (unlabeled SGI, data leak, complaint) triggers: (1) Immediate campaign halt, (2) Client notification "without undue delay", (3) Internal audit within 24 hours, (4) Report to Data Protection Board if DPDP breach (72-hour detailed report), (5) Platform compliance (3-hour removal for flagged SGI).
All team members trained on this policy at onboarding. Annual policy review or upon any new MeitY/IT Rules notification. Last reviewed: April 2026. Policy version: 1.0.