Using AI to write your marketing? Disclosure, fake reviews and ‘AI-powered’ claims

Do you have to disclose that AI wrote your marketing?

In most cases, no. No major market requires a label on a blog post, email or product description simply because AI helped write it. The general rule everywhere is the same: advertising must not mislead. If the copy is accurate and a person has checked it, the tool behind it rarely matters.

There are exceptions. In the EU, Article 50 of the AI Act has applied since 2 August 2026. It requires ‘deployers’, meaning businesses using an AI system, to disclose deepfakes, which are realistic AI images, audio or video of people, places or events. It also covers AI-generated text published to inform the public on matters of public interest, unless a human has reviewed it and someone takes editorial responsibility. Ordinary product copy is usually outside that text rule. A realistic AI ‘customer’ in a video ad may not be.

In the US, New York now requires advertisers to disclose ‘synthetic performers’, meaning AI-generated people who look like real actors, in ads shown to New York audiences. That law took effect on 9 June 2026, with penalties of $1,000 for a first breach and $5,000 after that. California’s bot law has, since 2019, made it unlawful to use an undisclosed bot to mislead people in California into a purchase.

Elsewhere, labelling duties mostly fall on AI providers and platforms rather than advertisers. China’s labelling rules (since September 2025), India’s IT Rules amendments (since February 2026) and South Korea’s AI Basic Act (since January 2026) all fit this pattern. If you use those platforms, expect their labels to appear on your content.

What do Google, Meta, YouTube and TikTok say about AI content?

Platforms move faster than lawmakers, and their rules can cost you reach even when no law is broken. Google’s position, published in February 2023, is that it rewards helpful content however it is produced. AI-written pages are not penalised just for being AI-written.

What Google does penalise is ‘scaled content abuse’, a spam policy added in March 2024. It targets generating many pages mainly to manipulate rankings, and it names using generative AI to produce pages ‘without adding value for users’ as an example. A site of 500 AI-spun location pages is the classic risk.

The social platforms focus on realistic content. Since March 2024, YouTube has asked creators to disclose realistic altered or synthetic content, while saying that AI help with scripts or captions does not need disclosure. TikTok requires labels on realistic AI-generated images, audio and video, and Meta adds ‘AI info’ labels where it detects or is told about AI. Failing to label can lead to removal or demonetisation.

Can you use AI to write reviews or testimonials?

No, not if they are presented as real customer opinions. This is the clearest rule in AI marketing. The US Federal Trade Commission (FTC) rule on consumer reviews and testimonials took effect on 21 October 2024. It bans fake reviews from people who do not exist or have no real experience of the product, and the FTC said expressly that this includes AI-generated fake reviews.

The same rule bans buying positive reviews, paying only for good reviews, hiding insider reviews, suppressing bad reviews through threats, and buying fake followers or views. Breaches can lead to civil penalties of up to $53,088 per violation. The FTC kept that figure unchanged for 2026. In December 2025 it sent the first warning letters under the rule to around ten companies.

The UK and EU go just as far. Since 6 April 2025, the Digital Markets, Competition and Consumers Act 2024 (DMCC Act) has listed fake reviews, and commissioning them, as ‘banned practices’, which are always unfair. The EU’s Omnibus Directive did the same from May 2022. Using AI to tidy the wording of a genuine review, with the customer’s agreement, is different. Inventing one is not.

What is AI-washing, and can you say your product is ‘AI-powered’?

‘AI-washing’ means overstating how much AI a product uses, or what that AI can do. It is the AI version of greenwashing. An ‘AI-powered’ badge is a factual claim, so you need evidence that it is true and that the AI does what you say.

Regulators have acted on this in several markets. In March 2024 the US Securities and Exchange Commission (SEC) fined two investment advisers, Delphia and Global Predictions, a total of $400,000 for false claims about their use of AI. In September 2024 the FTC’s ‘Operation AI Comply’ included DoNotPay, which paid $193,000 over its claim to be ‘the world’s first robot lawyer’. In August 2025 the FTC finalised an order stopping Workado from claiming its AI content detector was highly accurate without evidence.

In the UK, the Committee of Advertising Practice warned in 2024 that using AI to build a product is not the same as the product containing AI. In August 2026 the Advertising Standards Authority (ASA) summed up its approach to AI ads as ‘what the ad promises, the product should deliver’. It also warned against unrealistic earnings claims and against presenting AI as a replacement for a therapist.

• Say what the AI actually does, for example ‘AI suggests three subject lines’.
• Do not call a rules-based tool or a human team ‘AI’.
• Keep test results that back up any accuracy or time-saving figure.
• Do not claim AI beats competitors unless you have compared them fairly.
• Check claims your AI vendor makes before repeating them; you are responsible for your ads.

Do AI influencers and avatars need to disclose ads?

Yes. An AI influencer is treated like a human one for advertising purposes. The FTC’s Endorsement Guides, revised in June 2023, expressly cover virtual influencers. If a brand pays for or controls a post, the post must make the commercial link clear.

The bigger problem is honesty about experience. A virtual influencer cannot taste a protein shake or feel a moisturiser work. Presenting an AI avatar as a real user who has tried the product risks being a fake testimonial, which the rules above ban.

Other markets point the same way. The UK’s CAP Code requires ads to be obviously identifiable as ads, and the EU bans disguised advertising under its unfair commercial practices rules. India’s consumer regulator issued influencer guidance in January 2023, and its November 2023 dark patterns guidelines list ‘disguised advertisement’ as a banned dark pattern.

Can AI write and send your emails, texts and sales calls?

AI can write the message, but the usual consent rules still decide whether you may send it. In the US, CAN-SPAM requires honest headers and subject lines, your postal address and a working opt-out in commercial emails. In the UK, the Privacy and Electronic Communications Regulations (PECR) generally require consent for marketing emails and texts to individuals, with a ‘soft opt-in’ for existing customers. In the EU, GDPR and national e-privacy laws play the same role.

UK fines for e-marketing breaches rose sharply in 2026. Under the Data (Use and Access) Act 2025, PECR fines can now reach £17.5 million or 4% of global turnover, up from £500,000.

AI voice calls are the highest-risk area. In February 2024 the US Federal Communications Commission ruled that AI-generated voices count as ‘artificial’ voices under the Telephone Consumer Protection Act (TCPA). Calls using them need the same prior consent as any robocall, and people can sue for $500 to $1,500 per call.

Who is responsible if AI copy gets the facts wrong?

You are. Regulators do not accept ‘the AI wrote it’ as a defence. If an AI tool invents a statistic, a certification or a customer number and you publish it, it is your misleading claim.

AI tools also tend to write in confident superlatives. Words like ‘clinically proven’, ‘number one’, ‘cheapest’ or ‘cures’ need evidence before you publish, and health, finance and environmental claims face extra scrutiny everywhere. Comparative claims about named competitors must be fair, verifiable and like for like.

Copyright is a smaller but real risk. Purely AI-generated text and images may have no copyright protection in some countries, including the US, so competitors may be free to copy them. AI output can also occasionally reproduce someone else’s wording or style closely, so run important copy through a plagiarism check and avoid prompts that name living artists or rival brands.

What is happening in the market right now?

Enforcement against fake reviews has moved from guidance to action. In the UK, the CMA sent 54 advisory letters about fake reviews in July 2025 and, in March 2026, opened its first formal fake-review investigations under the DMCC Act. In the US, the FTC’s December 2025 warning letters were the first use of its reviews rule, and further cases are widely expected.

AI-washing cases have continued, but the focus has shifted. Since 2025 the FTC has concentrated on clear consumer harm, especially get-rich-quick schemes sold as ‘AI’. In January 2026 it announced settlements with Growth Cave, which it said falsely marketed ‘AI software’ to automate online businesses. In March 2026 it announced a proposed settlement with Air AI, banning it from selling business opportunities and including an $18 million judgment, largely suspended.

Meanwhile, disclosure duties are spreading through platforms and specific laws rather than general labels on all AI content. The EU’s Article 50 rules, New York’s synthetic performer law and the major platforms’ labelling policies all target realistic content that could deceive. Expect that direction to continue.