There was a time when beauty advice came from the trusted person behind the department store counter or from glossy magazines. Then came blogs. Then influencers. At every stage, consumers have relied on others to shape the conversation. Today, artificial intelligence (AI) is shifting that balance. Virtual try-ons, skin diagnostics, hyper-personalised routines, and automated shade matching promise to put agency in the hands of the user. But there’s a challenge behind convenience: What if AI gets beauty wrong?

In Asia Pacific, the margin for error is thin. It’s the most diverse, and fastest-growing beauty market in the world, on track to drive almost two-thirds of global beauty sales growth by 2027^.1 Consumers in the region look for experiences that reflect who they are, how they live, and what they value. Increasingly, savvy consumers, from Gen Z in Seoul to eco-conscious buyers in Sydney, expect a personalised, inclusive, and ethical beauty experience. But if AI is misused or misunderstood by brands, it can erode trust, alienate customers and create new forms of exclusion.

How to get it right? In this article, we’ll explore three key challenges AI brings to the beauty industry – and how bold, responsible brands can solve them.

1. Fixing flawed data and biased recommendations
   
2. Avoiding the trap of choice overload
3. Building trust in a privacy-conscious, emotionally driven category

The brands that get this right won’t just lead the beauty market. As PwC’s Value in Motion research reveals, the next decade is one where AI – along with climate change and shifting geopolitics – are reshaping how we live, work, and grow. They’re creating new customer needs, fuelling new business models, and blurring the boundaries of sectors and industries. 

We’ve seen it already. Once focused purely on cosmetics and skincare, today’s beauty brands from L’Oréal to Shiseido, are becoming platforms and more – branching into health, wellness, tech, and even finance. The opportunity is clear. So is the responsibility. Get it right, and beauty brands will do more than shape beauty trends – they’ll shape lives, industries, and entire ecosystems.

Three challenges, three imperatives 

1. Fix the bias problem: Precision is the new personalisation

'Bias’ in AI isn’t always deliberate – it’s a reflection of incomplete or skewed data. Many beauty algorithms were built on Western-centric datasets. Often, these fail to reflect Asia’s vast diversity of skin tones, undertones, facial features, and or cultural norms. The result: recommendation engines that misread melanin, shade matchers that exclude mid-to-deep tones, or virtual try-ons that render inaccurately under Asian lighting conditions. This isn’t only a technical flaw, it’s a commercial and cultural risk. Exclusion erodes brand equity, undermines trust, and leaves money on the table – especially in growing markets such as India, Indonesia, Vietnam, and the Philippines.

What brands can do:

• Localise data at the source. Build representative datasets that reflect the region’s diversity: full ranges of skin tones and undertones; humidity, UV (Ultraviolet radiation), and pollution by city; cultural preferences by market; ingredient sensitivities (such as fragrance, alcohol, and halal requirements). Partner with dermatologists and local creators to creators to provide commentary on nuanced attributes.
• Co-create data with customers. Follow innovators like Amorepacific, whose in-store AI-powered beauty labs use real-time consumer inputs with robotics to custom-mix products on the spot. While advanced algorithms recommend personalised lipstick shades.² Every interaction becomes a data point that improves the model and product.
• Detect micro-trends, not just macro-ones. ‘Glass skin’ in Korea, ‘sun-kissed glow’ in the Philippines, and a ‘no-makeup’ aesthetic in Japan. These are actionable signals for tailoring recommendations, content, and even R&D - to go beyond marketing. 
• Put humans in the loop – beauty advisors, creators and dermatologists who can validate outputs and challenge the model. When model confidence is low (for example, in a shade match), hand-off to a person, seamlessly.
• Make it a community effort – invite consumers to flag issues, contribute images under varied lighting, and co-create inclusive experiences. Reward participation to deepen engagement and improve coverage.
• Be ‘radically transparent’ – publish model cards that state what the model does, where it performs well and poorly, and how accuracy and fairness are improving. Share bias metrics and action plans openly.
• Audit continuously. Test for disparate performance across skin tones, genders, and age groups. Red team your algorithms to probe failure modes. Commission independent audits where credible.

2. Avoid the choice overload trap: Simplify or lose customers

More AI doesn’t always mean better decisions. In many Asia Pacific markets, shoppers are already facing a flood of SKUs (stock-keeping units), product recommendations, reviews, and hyper-personalised experiences. AI that simply adds more options, risks creating confusion rather than clarity. This is especially true in Southeast Asia. Beauty ideals and routines vary widely by culture, climate, and skin type. Live commerce and social platforms amplify product proliferation.

The result? A growing ‘paradox of choice’ where too many personalised options actually hinder decision-making and leave consumers feeling less satisfied, instead of more.

What brands can do:

• Prioritise relevance over volume – cap recommendations and design for fewer, better-curated options based on real-world context, not merely browsing history. Take L’Oréal’s Beauty Genius AI assistant – it simplifies solutions via curated, dermatologist trained diagnostics, and virtual try-ons across 750+ products.³ Simplicity builds confidence. Confidence builds conversion.
• Explain the why – opaque suggestions erode trust. A concise rationale – ‘Humidity will be high in Manila this week; this oil-control formula tested best for similar profiles’ – reduces cognitive load.
• Anchor on jobs to be done and emotion. Start with what the user is trying to achieve –sensitivity, hyperpigmentation, time constraints – or how they want to feel and show up. In Asia’s beauty culture, ritual matters as much as results: ‘Skincare to unwind after work,’ ‘soft morning look,’ or ‘bold date night’ can be more resonant than ‘anti-aging serum.’
• Use constraints intentionally. Default to full-size plus refill rather than three new products; promote multi-use and refillable items where appropriate. Our research suggests sustainability is a growing purchase driver; AI that supports thoughtful use – for example routines built around what customers already own – earns loyalty.
• Close the loop with assortment. Feed recommendation data into SKU rationalisation. Prune long tails that create confusion without adding value. Reinvest in hero products and formats that travel well across climates and cultures. 

3. Earn and keep trust: AI must be responsible by design

Personalised skincare and cosmetic recommendations often rely on deeply sensitive data – face scans, skin condition, behavioural patterns. But many beauty brands still treat AI as a black box. They collect data without clear consent and offer recommendations without explaining how or why. Not only is this a compliance risk – it’s a business risk.

Asia Pacific consumers are privacy-conscious and digitally savvy. Trust has never been more critical – or fragile. Seven out of ten people express concerns about how their information is being collected by companies. That figure climbs even higher in the Philippines (86%), Thailand (81%), and Singapore (81%).⁴

And expectations are rising. Regional privacy laws – like China’s Personal Information Protection Law,⁵ Singapore’s Personal Data Protection Act,⁶ and India’s emerging Digital Personal Data Protection Act ⁷ – are putting more pressure on brands to be transparent, responsible, and proactive. 

What brands can do:

• Adopt ‘radical transparency’ as a competitive advantage, not a checkbox. Be up-front about when you are using AI – since it is not always clear. It’s especially important in a category as universal and emotionally resonant as beauty. Communicate when synthetic media is used. Avoid deepfake influencers unless clearly fictional. Ensure diversity in both human and AI-generated visuals. This transparency extends to the data itself – explain what is being collected, how it will be used, and why it matters. One approach can be to use zero-party data – information that customers willingly and directly share with a brand. In a space that spans generations and identities, transparency builds trust. And trust builds loyalty.
• Minimise and protect. Collect only what is necessary. Where possible, process sensitive data (such as face scans) on-device. Use techniques like federated learning and differential privacy to reduce central data risks. Encrypt by default; segment access tightly.
• Go beyond the fine print, offer clear opt-outs and allow users to adjust or decline recommendations. Let them feel in control. 
• Keep models fresh and feedback-driven – beauty moves fast. Trends shift. Seasons change. Skin needs evolve. A model trained last year might already be outdated or biased. Continuously refine AI based on new data and user feedback – and let customers know you’re doing it.
• Treat Responsible AI (RAI) overall as a differentiator – in a category built on confidence, trust is everything. Don’t just meet regulations, set the standard for the industry. Build Responsible AI principles and a framework into your ways of working. That could mean standing up a dedicated, cross-functional RAI board with the authority to assess bias, review potential harm, approve launches, and oversee model updates. Make responsible AI part of your brand DNA, not only a compliance exercise.

How else might AI reshape the beauty industry?

One of the biggest shifts ahead won’t necessarily come from within the beauty industry – but from outside it. As our Value in Motion research shows, new entrants are emerging from adjacent sectors. Digital-first, nimble, and often purpose-led, they’re building new kinds of customer connection – powered by smarter data, AI, and agile business models. Just like the microbeauty brands that have already reshaped the market, these newcomers won’t play by the old rules – and they’ll win market share fast.

The opportunity for large beauty brands? To do the same. To evolve business models and capitalise across these new growth domains. Using AI to drive efficiency, and to capture sharper consumer insights to fuel new products, services, or even new businesses. It means looking beyond traditional sector and category lines, and partnering across industries – where beauty intersects with wellness and performance, healthcare, wearables, and more. This is beauty redefined – not solely by how you look, but also how you feel and function.

So, what practically might that future look like?

• Smarter insights, deeper connection and personalisation – AI is unlocking more than preferences alone. It’s revealing emotion, identity, and intention too. Facial scanning, eye tracking, and real-time mood detection can help brands understand what consumers buy, and how products make them feel. This unlocks personalisation at scale: hyper-targeted products, diagnostics, and experiences tailored to individual needs. Shiseido, with its diagnostic ‘Skin Visualizer’ is making bold moves in this space.⁸ And in the right categories, consumers may be willing to pay a premium for it.
• From reactive to predictive – next-generation AI won’t wait for a click. It will anticipate needs based on mood tracking, voice tone, calendar data, or even sleep quality. It will serve up proactive beauty solutions before the customer asks.
• Agentic AI that acts – virtual assistants are gaining traction across other industries. In beauty, it could mean assistants that do more than suggest products – they act. Scheduling treatments, reordering refills, adjusting skincare based on the weather or stress.
• Devices and wearables. Partnerships with device makers – smart mirrors, cleansing tools, UV sensors – enable on-device analytics that preserve privacy while deepening personalisation. Retail becomes phygital: live diagnostics in store, continuity at home.
• Ecosystems over SKUs. Beauty-as-platform connects products to services: teledermatology, nutrition, sleep coaching, mental wellness. Payment and financing can integrate with superapps and regional wallets, smoothing cross-border commerce. 
• Scale is an advantage, if used well. Yet big beauty brands still struggle to beat microbrands at their own game. Prestige and global appeal no longer guarantee cut-through. Today, local and digital-native brands feel more personal – and more relevant. But big brands can out-invest, out-analyse, and out-innovate. But that only works if they use their scale to move up the tech curve – not stay stuck in legacy systems. 
• Modernise the stack (cloud, feature stores, machine learning operations), consolidate and clean data. And partner with regional platforms (from e-commerce giants to chat superapps) to meet consumers where they already are. The AI talent market is tightening – remember those wage premiums and growth in AI-intensive roles flagged by our AI Jobs Barometer – so build, buy, and partner with intent.

The future belongs to bold, responsible brands

Beauty brands in Asia Pacific are at a crossroads. Get AI right, and the upside is huge: more relevance, stronger loyalty, sustained growth. Get it wrong – through bias, choice overload, or misused data – and risk undermining progress. Beauty brands could lose trust, loyalty and cultural relevance – everything brands have worked so hard to build.

But within those risks lie the biggest opportunities. With the right data, governance, and intent, AI can unlock deeper personalisation, enable bold cross-industry moves, and push beauty into entirely new domains of growth. The brands that move early won’t just define the future of beauty. They’ll expand what beauty can be.

Published on 8 July 2025