Introduction: “Possible” Aur “Publishable” Same Nahi Hain
Generative AI ab kisi famous face, voice aur mannerism ka convincing synthetic performance create kar sakta hai. Production capability exciting hai, but advertising mein technical possibility only first gate hai. Consent kisne diya? Source material ka right kis ke paas hai? Audience ko kya disclose hoga? Brand backlash ka owner kaun hoga? Agar person deceased hai, legacy custodians and family expectations kaise handle honge?
SKF ke Greta Garbo campaign ne in questions ko sharply visible banaya. Reuters ne Oct 2, 2026 ko report kiya that the Swedish bearings company used multiple AI tools to recreate Garbo for an advertisement. SKF said the production was developed with her estate and family; the voice used authorised material from an early film, while the visual likeness was generated through text prompts rather than archival photos or footage because of copyright constraints.
Indian marketers ke liye lesson “AI celebrity ads banaiye” ya “never use them” nahi hai. Lesson yeh hai ki synthetic likeness campaign ko normal creative asset nahi, high-risk rights and reputation project treat karna chahiye.
SKF–Greta Garbo Case Mein Kya Confirm Hua?
Garbo, who died in 1990, had appeared in a 1921 promotional film involving SKF early in her career. The new campaign deliberately used that historical connection. Reuters reported that image generation involved ByteDance’s Seedream 5 Pro and Google’s Gemini Nano Banana Pro; motion work used ByteDance’s Seedance models and Kuaishou’s Kling AI. A human actor helped with pacing and cadence.
The Guardian reported on Sep 30, 2026 that Garbo’s relatives who manage her legacy were involved and that the 99-second advertisement was her first posthumous moving-image appearance. The campaign also drew criticism, showing that rights clearance does not automatically produce cultural approval.
Confirmed facts
- SKF commissioned an AI-generated portrayal of Greta Garbo for advertising.
- The company says it collaborated with Garbo’s estate and family.
- Reuters says authorised audio from one early film was used for the voice; visual generation did not rely on archival Garbo imagery.
- Multiple U.S. and Chinese AI tools plus human performance input were used.
- The campaign attracted both praise and criticism.
What remains context-dependent
- Public reporting does not reveal every contract term, territory, duration or approval right.
- Estate cooperation does not prove that every copyright, trademark, performer or publicity issue is identical across jurisdictions.
- A technically impressive result is not evidence that consumers will trust or like it.
- This case does not create a universal legal template for India.
Likeness, Voice Aur Copyright Ko Ek Cheez Mat Samjhiye
AI campaign teams often ask, “Do we have permission?” But permission ek single checkbox nahi hai. At least four different layers evaluate karni chahiye.
1. Person or estate authorisation
A living celebrity can approve use through a contract; a deceased personality may have estate, family or appointed legacy custodians. Exact authority and enforceability territory-specific ho sakte hain. Verbal enthusiasm or public tribute legal clearance ka substitute nahi hai.
2. Source-material rights
Even when a personality approves, photographs, film clips, sound recordings, scripts, costumes, trademarks and music can have different rightsholders. SKF case demonstrates this distinction: reported visual generation and authorised voice-source treatment were not identical.
3. Output and model terms
Agency ko each tool’s commercial-use terms, training restrictions, indemnity position, content policy and output ownership review karna chahiye. “Enterprise plan” label automatically campaign-specific legal safety guarantee nahi deta.
4. Advertising and consumer interpretation
Audience ko lag sakta hai ki celebrity ne product personally endorse kiya. Disclosure, context and creative framing therefore material hain. India mein campaign release se pehle legal counsel and applicable advertising standards review essential hai; case-specific legal requirements qualified counsel se confirm karein.
India-Specific Brand Implications
Celebrity culture amplifies both reach and risk
Bollywood, cricket, music and creator-led marketing India mein high attention command karte hain. A recognisable synthetic likeness campaign quickly travel kar sakta hai—but parody, impersonation, misleading endorsement and emotional exploitation accusations bhi equally fast spread ho sakte hain.
Deceased personalities need higher ethical scrutiny
Legal clearance minimum threshold hai. Ask whether the person had publicly stated preferences, whether the performance aligns with their known values, and whether the commercial category creates discomfort. Family approval valuable hai, but audience may still disagree with the use.
Voice cloning has special fraud adjacency
Synthetic voice advertising and scam calls share similar consumer cues. A campaign that normalises an unauthenticated celebrity voice can unintentionally weaken trust. Watermarking, disclosure, owned-channel verification and rapid takedown process therefore brand-safety controls hain, not optional polish.
Regional languages increase review complexity
If likeness speaks Hindi, Tamil, Bengali, Marathi or another language, literal translation enough nahi. Accent, idiom, cultural meaning, claims and pronunciation need native review. Approval should cover each language version, not just the master English film.
Campaign Approval Ke Liye Eight-Gate Framework
Gate 1: Strategic necessity
Why this person? Agar campaign idea only shock or nostalgia par depend hai, long-term value weak ho sakti hai. Historical or product connection—like SKF’s reported link to Garbo—creative rationale strengthen kar sakta hai, but does not remove other risks.
Gate 2: Authority map
Living person, estate, family, agent, studio, photographer, record label, writer and brand owners ka rights map banaiye. Unknown authority ko “probably fine” mark na karein. Jurisdiction, media, duration, edit rights, model-training use and future reuse separately document karein.
Gate 3: Source ledger
Every reference image, audio clip, film frame, script line, costume detail and logo ka source and licence record rakhein. Prompt-generated visual bhi inspiration chain se free nahi hota; legal review ke liye reproducible asset history useful hai.
Gate 4: Model and vendor review
Tools, versions, account tier, uploaded inputs, retention settings, subcontractors and geographic processing record karein. Vendor ko deletion, model training, confidentiality and incident notification questions ask karein.
Gate 5: Human creative accountability
AI ko director credit dekar responsibility transfer nahi hoti. Named executive producer, legal approver, brand owner and ethics reviewer assign karein. Human performance capture use ho to actor contract and consent clearly handle karein.
Gate 6: Disclosure design
Disclosure visible, understandable and medium-appropriate honi chahiye. Tiny end-frame text ya hidden caption consumer confusion prevent nahi kar sakta. Campaign, landing page, social post and press material par consistent language use karein.
Gate 7: Pre-release stress test
Small diverse group ko creative dikhaiye and ask what they believe: Is the person alive? Did they record this? Are they endorsing the product? Misinterpretation rate high ho to context or disclosure improve karein. Sentiment testing legal opinion ka replacement nahi, but reputation risk surface karta hai.
Gate 8: Launch and takedown plan
Monitor impersonation copies, misinformation, offensive remixes and scam reuse. Platform reporting contacts, asset fingerprints, response statement and decision authority ready rakhein. A pre-agreed kill switch campaign team ko hours save kar sakta hai.
Creative Brief Mein Kya Likha Hona Chahiye
- Business objective and why the specific personality is relevant.
- Exact approved likeness, voice, expressions, wardrobe and prohibited contexts.
- Source assets and permitted transformations.
- Markets, languages, channels, duration and paid-media rights.
- Disclosure wording and placement.
- Product claims and substantiation owner.
- Human performer involvement and credit.
- Approval rounds, final sign-off and change-control rules.
- Data retention, vendor deletion and future model-training restrictions.
- Complaint, takedown and crisis-response workflow.
AI-assisted campaign assets ke liye creative design services ka brief approved likeness rights, disclosure wording aur final creative review ke saath prepare karein.
30-Day Pilot Plan for Indian Brands
Days 1–7: Decide whether AI likeness is necessary
Alternative routes compare karein: archival licence, living actor, animation, fictional character or voiceover. Expected attention ko rights cost and reputational risk ke against evaluate karein. Board-level visibility appropriate ho sakti hai for high-profile personalities.
Days 8–14: Complete rights and vendor diligence
Written permissions, source licences and territorial scope obtain karein. External counsel se relevant Indian and campaign-market issues review karayein. Tool terms and data handling document karein.
Days 15–21: Produce a limited proof
Public media buy se pehle short internal cut banaiye. Lip sync, facial artefacts, language accuracy, claims, disclosure and accessibility test karein. All prompts and approved output versions maintain karein.
Days 22–30: Stress test and approve
Audience interpretation test, crisis simulation and impersonation-monitoring plan run karein. Final asset ko named legal, brand and executive approvers sign off karein. Unresolved objection ho to launch delay or concept change should remain a valid option.
Confirmed Fact vs Professional Analysis
Confirmed fact: SKF used generative AI and human input to create a Garbo advertisement; Reuters reported estate and family collaboration, authorised voice-source use, and no archival visual input for the recreated likeness. The campaign was publicly debated.
Indian brands should treat synthetic celebrity work as a high-risk governance project with layered permissions, disclosure, provenance and response controls. The eight-gate framework is practical risk guidance, not a statement that any specific campaign is legally compliant or non-compliant.
Conclusion
AI can make a historic personality appear to speak again, but it cannot answer the core business questions: Should the brand do it? Who has authorised it? Will consumers understand what they are seeing? Can the team defend every source, edit and claim?
The strongest campaign will not be the one with the most realistic face. It will be the one with a clear strategic reason, documented rights, respectful execution, transparent disclosure and a tested exit plan.
Plan Responsible AI Creative for Your Brand
AI creative, brand approvals aur campaign governance ka roadmap chahiye? Hemant Kumar Sharma ke saath AI marketing consultation explore karein. Call/WhatsApp: +91 98116 81687.
Sources
- Reuters, “Swedish ball-bearing ad revives Greta Garbo with US, Chinese AI” — published Oct 2, 2026: https://www.reuters.com/business/media-telecom/swedish-ball-bearing-ad-revives-greta-garbo-with-us-chinese-ai-2026-10-02/
- The Guardian, “Reanimated AI Greta Garbo stars again … in a ball-bearing advert” — published Sep 30, 2026: https://www.theguardian.com/film/2026/sep/30/ai-greta-garbo-stars-again-ball-bearing-advert
- SKF campaign video, “Pay Like It’s 1919 — With Your Energy Savings. Starring Greta Garbo.” — published Sep 30, 2026: https://www.youtube.com/watch?v=JbttlHN0A4U
ASCI Final AI-Ad Labelling Rules: Brand Checklist
ASCI released final Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising on Sep 29, 2026, following the draft consultation opened in May. The guidelines take effect three months from publication and apply a consumer-impact test: disclosure is required when synthetic content materially influences consumer understanding or decision-making and hiding its origin could mislead.
This directly strengthens the governance framework above. Estate or celebrity consent may solve one permission layer, but consent does not remove the need for an AI label where the synthetic performance materially influences the audience.
Mandatory labelling examples
- Synthetically generated influencers and brand ambassadors.
- Replication of a real person’s likeness or voice—with consent—for personalised messaging.
- Fabricated events or settings that can affect understanding of the advertised product or service.
- Demonstrations involving products that do not currently exist.
- Synthetic sound effects closely connected to a product’s core feature, such as headphones.
- Paid or sponsored AI-generated product recommendations; ASCI specifies “Sponsored by [Brand]”.
Uses where ASCI says labelling is not required
- Routine colour, light, noise or minor-blemish correction.
- Decorative or ambient elements.
- Obviously fantastical effects.
- Administrative and text-based use, including advertising-copy generation.
- Accessibility uses such as subtitles, translations and accurate audio descriptions.
A label does not cure prohibited content
ASCI says fabricated endorsements, misleading product-result exaggeration, deceptive presentation of non-existent locations, unauthorised copyrighted work, and deepfakes or likeness use without consent remain prohibited. “Created using AI” cannot make an otherwise misleading advertisement acceptable.
Practical implementation before the transition ends
- Live and scheduled campaigns ka synthetic-content inventory banaiye.
- Each asset ko prohibited, mandatory-label, or low-risk/no-label category mein classify karein.
- Video, audio, static, influencer and personalised variants separately review karein.
- Platform label available ho to bhi wording accuracy and visibility verify karein.
- Disclosure ko ASCI disclaimer requirements ke saath align karein.
- Agency, influencer and production contracts mein synthetic-media declaration add karein.
- Rights, consent, source assets, prompts, tools and final approvals ka evidence retain karein.
- Ambiguous campaign ko publication se pehle legal and ASCI-focused review dein.
ASCI’s final press release dated Sep 29, 2026 sets out a three-month transition, mandatory and non-mandatory examples, and prohibited uses. Check your campaign and contracts against that guidance, and seek a formal ASCI opinion or legal advice where needed.
Sources:
- ASCI, “Guidelines for Responsible Labelling of Synthetically Generated Content in Advertising” — final press release published Sep 29, 2026: https://www.ascionline.in/wp-content/uploads/2026/09/Press-Release_Guidelines-for-Responsible-Labelling-of-Synthetically-Generated-Content-in-Advertising.pdf
- ASCI Code for Self-Regulation — current reference checked Oct 5, 2026: https://www.ascionline.in/the-asci-code/