Artificial testimonials and their impact on trust and compliance

I’ve observed a dramatic rise in brands using AI-generated avatars as artificial customers in advertising. This is a problem: it negatively impacts credibility and erodes trust. Presenting a fabricated customer experience as genuine also risks breaching Australian Consumer Law.

The issue is straightforward. When a testimonial is delivered by an artificially generated character, my concern is that consumers will assume the entire customer experience is fabricated or artificial. That can undermine confidence in the message. My commercial concern is the effect on persuasion, response and acquisition costs over time. A clearly identified synthetic presenter conveying a real, accurately represented experience is a different case from inventing a customer and their endorsement.

Under Australian Consumer Law, the rules are clear and apply regardless of production method. Section 18 prohibits misleading or deceptive conduct. Section 29 prohibits false or misleading representations about testimonials, product characteristics, or the nature of the experience being promoted.

A clear enforcement example is the Service Seeking case (2020), where the Federal Court issued $600,000 in penalties after the platform allowed businesses to publish self-authored reviews that were presented as genuine customer testimonials. The conduct breached Section 18 and Section 29. This is not an isolated action. Regulators have consistently penalised fabricated testimonial practices, including actions against P & N/Worldwide Energy (2014), Citymove (2011), and several solar providers. These cases confirm that misrepresenting customer experience, whether manually produced or AI-generated, sits squarely within established enforcement boundaries.

At the platform level, Google, Meta and TikTok have rules against misleading content. AI disclosure requirements differ by platform and format; they are not a universal rule that every synthetic presenter is prohibited. An AI label does not make a fabricated endorsement truthful.

Despite these policies, the tactic is proliferating at speed. A recent example circulating on Facebook shows an individual using Arcad to generate a fabricated testimonial for Lululemon. To be clear, this is not Lululemon running the ad. The point is that an entire cottage industry has emerged to produce artificial reviews, influencer-style videos, and synthetic promotional content. The incentives are clear: it is fast, cheap, and scalable, with real compliance and trust problems when the experience being sold is fabricated.

To illustrate how widespread this is becoming, a search for “AI Avatar” in the Meta Ad Library reveals hundreds of ads promoting AI-generated presenters, avatar-based endorsement tools, and tutorials teaching marketers how to manufacture “customer” videos at scale. While these listings do not show the testimonial videos directly, they show how readily available the production tools have become. They do not establish how many advertisers are publishing false testimonials.

Accountability sits with both marketers and platforms. Brands and agencies are responsible for ensuring testimonial content is truthful, compliant, and based on genuine customer experience. Platforms, however, hold structural responsibility. Watermarking technologies such as Google DeepMind’s SynthID can help identify content carrying their watermark. They cannot detect every synthetic image or video, or establish whether the customer experience described is true. In my view, platforms should combine provenance tools with review and enforcement of misleading-testimonial rules.

The consequences extend beyond monetary penalties. Misleading testimonial practices damage long-term brand equity, undermine future campaign performance, and create systemic distrust across an entire category. Effective marketing relies on credibility. Artificial customers, when used in testimonial contexts, degrade that credibility and introduce unnecessary legal and commercial risk.

The principle is straightforward. AI has an important role in production, optimisation, and workflow efficiency, but it cannot replace real customers in testimonial formats. Artificially generated personas should not be used to simulate authentic customer experience. The long-term deterioration in trust outweighs any short-term production gain.

Real customers build credibility. Artificial customers erode it. Sustainable marketing requires clarity on that distinction.

Regulation and penalties

Area	Law / Rule	Requirement	Penalties / Examples
Misleading conduct	ACL Section 18	Conduct must not mislead or deceive. AI-generated testimonials presented as real risk breaching this.	Injunctions and other remedies. Section 18 alone does not attract civil pecuniary penalties; Section 29 can.
False testimonials	ACL Section 29(1)(e),(f)	Prohibits false or misleading testimonial representations, including representations concerning a testimonial.	Service Seeking (2020): $600,000. P&N/Worldwide Energy (2014): $125,000. Citymove (2011): $6,600.
Advertising standards	AANA Code of Ethics, Section 1.2	Advertising must not be misleading or deceptive. Assess the representation and context, not simply the use of AI.	Self-regulatory complaints process; separate ACL enforcement may apply.
Platform compliance	Platform AI-disclosure rules	Follow the applicable format’s disclosure requirements and do not misrepresent customer experience.	Ad rejection, account penalties, campaign removal.

Platform policies

Platform	Policy requirement	Summary rule	Consequence
Google	Misrepresentation & review integrity	False or misleading endorsements can breach advertising rules; separate review-product policies also apply.	Ad or feed rejection; account warnings or suspension.
Meta	AI-content disclosure & misleading content	AI labelling depends on the product and context; misleading representations remain subject to advertising rules.	Ad rejection; forced labelling; account restrictions.
TikTok	AI-content labelling & synthetic persona rules	Apply the relevant synthetic-media labelling rules; avoid false or misleading endorsements.	Ad rejection; removal; penalties.