The new personalisation paradox: Can Meta balance AI chat relevance with privacy?

Recently, Meta has unveiled plans to enhance personalised user experiences by incorporating interactions with its generative AI (GenAI) features. Beginning 16 December this year, conversations with Meta AI—via text or voice—will influence the content and advertisements users encounter across platforms such as Facebook and Instagram.

According to Meta, the goal is to deliver more relevant recommendations by better understanding individual interests and behaviours. “Soon, interactions with AIs will be another signal we use to improve people’s experience,” the company shared in a blog post.

This push for personalisation aligns with modern consumer expectations. A McKinsey report highlights that 71% of consumers now expect companies to deliver tailored interactions, and 76% become frustrated when this doesn't happen.

Don’t miss: Meta to use AI chats to personalise ads and content across platforms

A new channel to understand consumer intent

With the upcoming shift toward AI interactions, new ways to understand consumer interests are being introduced, providing additional signals that help advertisers recognise intent more accurately, according to Kenzo Selby, managing director, Japan, GumGum.

AI can help marketers interpret what topics, products, and ideas capture people’s attention at scale, giving brands a clearer sense of real-time intent without compromising privacy, he said.

However, AI interactions should not be viewed as a replacement for first-party data, which is collected and managed directly by brands through their own channels with user consent, allowing for long-term accumulation and leverage, he added. 

First-party data includes transactional and demographic information, as well as web analytics, behavioural data, and implied interests, but AI interactions are simply another data source that falls under behavioral data and implied interests, according to Jessica Liu, principal analyst at Forrester. “Marketers will continue to identify consumer intent using whichever of the aforementioned data categories are available to them."

Is Meta's exclusion of sensitive topics enough for ethical targeting?

While Meta has clarified that sensitive topics—such as religion, health, and sexual orientation—will not be used for ad targeting, experts are skeptical about the effectiveness of this safeguard.

Stephanie Liu, senior analyst, Forrester pointed to Meta's track record, citing multiple accusations of allowing illegal ad targeting based on age, gender, and race. She warned that simply excluding obvious keywords is a minimal approach, as advertisers could easily use proxy variables to target the same groups, whether intentionally or not.

"Meta has faced multiple accusations of letting advertisers target ads illegally — things such as targeting job ads based on age or gender, or financial services ads based on race."

Back in July, a California jury has ruled against Meta in a privacy lawsuit concerning the alleged collection of sensitive user data from Flo, a period-tracking app. Flo Health assured users that their sensitive reproductive health data and responses to survey questions would remain private. However, that personal information was ultimately shared with companies such as Meta and Google through their respective advertising tools.

Excluding sensitive topics is a responsible starting point; however, ethical targeting requires a deeper understanding of the audience's mindset and environment, according to GumGum’s Selby. He cited the death of US political activist Charlie Kirk on 10 September as an example, noting that GumGum’s data reveals that attention on ads appearing on related pages dropped sharply to just 1.5 to two seconds, even as page views surged. 

This shows that ads placed beside polarising stories not only risk brand safety but also lose effectiveness, despite increased traffic.

Echoing his perspective, Stella Leung, SVP of Greater China and Korea at The Trade Desk, emphasised that AI interaction data is typically generated within large, closed platform ecosystems, making it inaccessible for brands to directly own or analyse. This creates a “black box” situation, complicating ethical considerations.

Moving forward, the crucial challenge for brands will be to responsibly integrate signals from these closed platforms with their own first-party data, as well as third-party and retail media data within the open internet ecosystem, all while respecting user privacy, according to Leung. 

How to balance AI-driven personalisation with consumer demands for privacy?

Consumers today value personalisation but expect it to be handled with care and respect for privacy. According to GumGum Japan’s Digital Advertising Pulse Check, 81% of Japanese consumers find it important to protect their personal data while receiving ads, and 37% pay more attention to ads relevant to their interests. 

However, many feel uneasy when ads appear overly personal; when confronted with ads that seem to know too much about their private lives, 31.9% become more determined to protect their data, while 17.9% question how brands obtained that information, according to the report. 

To meet these expectations, brands must establish a personalisation vision grounded in customer needs and mutual value exchange to inform product updates, such as those implemented by Meta, according to Forrester’s Jessica Liu. 

A significant technical hurdle, according to The Trade Desk's Leung, is that AI interaction data is often locked within "closed platform ecosystems," creating a "black box" for advertisers. She argued that the future challenge will be responsibly integrating signals from these closed platforms with a brand's own first-party data in the open internet, all while respecting user privacy.

Leung believes solutions such as Unified ID 2.0 demonstrate that personalized advertising can coexist with user control. The same principle, she said, must apply to AI: users should understand how their interaction data is used and have the ability to opt out.

To balance AI-driven personalisation with privacy expectations, advertisers must understand the right mindset and connect at the right moment. AI can facilitate this by combining insights about audience mindset and attention, enabling brands to deliver meaningful experiences that respect privacy while maintaining impact, said GumGum’s Kenzo Selby. 

When advertising aligns with what a person is actively thinking about, it becomes relevant without relying on personal or behavioural data.

Related articles:

Meta to use AI chats to personalise ads and content across platforms
Meta partnership powers SM's data-led future as new China mall nears opening
Meta joins SPF and NCPC to beat scammers at their own game