Where AI helps
There is one rigorous experiment on writing, and a few studies of freelance, stock-image and retail platforms. The rest are figures that companies and software vendors published about themselves.
| Work | What has been reported | Where it comes from |
|---|---|---|
| Copywriting | Professionals finished writing tasks 40% faster, with better quality. | Experiment with 453 professionals |
| Images, translation, routine copy | A client made images in 7 days instead of 6 weeks and cut spending on outside suppliers by 25%. | Klarna case |
| Product descriptions | The retailer Adore Me cut monthly time on marketplace product descriptions from 20 hours to 20 minutes. | A case study by the software vendor |
| Search advertising | Bayer in Australia used machine learning to predict demand and adjust its search ads. It reported an 85% higher click-through rate and a 33% lower cost per click, year on year. | A case study by Google |
| Freelance writing, translation and coding | On one large freelance platform, demand for writing and translation fell 20–50% below its earlier trend after ChatGPT, and short jobs fell most. On another study’s platform, posts for writing and coding jobs fell 21% within eight months. | Study of one freelance platform, a peer-reviewed study by Hannane, Demirci and Zhu |
| Stock images | After a large stock-image platform allowed labelled AI images, total image sales rose 39%. Sales of images not made with AI fell, and contributors who did not use AI left. | A working paper by Goldberg and Lam |
| Ad copy | On a large retail platform, AI-written Google ad titles showed no gain in sales. | Randomised experiments on one platform |
| An AI platform across a whole agency group | WPP spent £250 million in 2024 and about three in ten staff were active users. | WPP case |
Note who the first rows are about. They are clients and brands, not agencies. Much of the published gain went to the buyer of marketing work.
Risks and rules
- Clients take the work in-house. Klarna said it cut spending on translation, production, customer-relationship and social-media agencies by a quarter. WPP warned its own investors that clients can now license the same AI tools themselves. WPP invested heavily in AI and its revenue still fell in 2025. Smaller clients do the same: 18% of US small firms that use AI spent less on outside services, in a Federal Reserve survey.
- You may not own what the tool made. In the United States, the Copyright Office concluded in January 2025 that copyright “does not extend to purely AI-generated material”. It also concluded that, with today’s tools, prompts alone do not give a person enough control to count as the author. A person’s own contribution is protected: what they wrote or drew, and how they selected, arranged or changed the AI output. Each work is judged case by case. This matters when a client expects to own the work outright.
- Labelling AI-made content. In the EU, the AI Act says that whoever uses AI to make a deep fake must disclose that it is artificial. A deep fake is AI-made or AI-altered image, audio or video that looks like real people, places or events and would wrongly appear authentic. For work that is clearly creative, satirical or fictional, the disclosure may be made in a way that does not spoil the work. Separately, the makers of AI tools must mark their output so software can detect it. Tools already on the market before 2 August 2026 have until 2 December 2026 to do so.
- The advertiser answers for automated ads. In the United Kingdom, the advertising regulator says its code applies however an ad was made, and that automation “doesn’t transfer liability”. In an earlier ruling, Google’s automated tools had put together an ad for a small clothing retailer, and the regulator held that the retailer was still primarily responsible. If an ad platform’s AI builds ads for your client, someone must check them.
- Fake reviews and testimonials. In the United States, a federal rule announced in August 2024 bans fake reviews and testimonials. The regulator named AI-generated fake reviews. The rule also bans buying fake followers or views.
Practices for this industry
All 30 practices in the library apply in this industry. None of them is specific to one sector. For the ones below, a study, a company case or a worked example from this sector shows why they matter here.
- Ask AI vendors for proof before you buy, and test on your own material An AI-text detector advertised as 98% accurate tested at 53% on general content, according to the US consumer regulator.
- Measure time saved per task against a baseline Klarna, a client of agencies, published before-and-after figures for image production: 6 weeks down to 7 days.
- Claim only what your AI has been tested to do Agencies write the claims. One vendor paid $1 million over a promise that its AI plug-in made any website compliant, and over reviews that were paid for.
Pick your company size
Size changes what is practical more than sector does. Each link opens the library filtered to that size.
Company cases
Each case is a named company with numbers and a public source. All cases.
- accessiBe: an AI plug-in sold as making any website accessible, and a $1 million order accessiBe sold website owners an AI plug-in and said it would make any site meet accessibility standards. The US consumer regulator said it did not, and that reviews praising it were paid for. The company paid $1 million.
- Klarna: AI cut spending on outside marketing suppliers by a quarter Klarna reported that AI tools let it spend 25% less on outside marketing suppliers while running more campaigns. For agencies, it is a client saying in public which work it has taken in-house.
- Workado: an AI detector sold as 98% accurate that tested at 53% Workado sold a tool that claims to tell AI-written text from human writing, and advertised it as 98% accurate. The US consumer regulator said independent testing put it at 53% on general text. Buyers had only the vendor’s word.
- WPP: £300 million a year on an AI platform, and revenue still fell WPP, one of the largest advertising groups, invested heavily in its own AI platform and reported wide use by staff. Its revenue still fell in 2025. Investing in AI did not, on its own, protect an agency’s income.
Studies from this sector
These studies were run on work that is typical of this sector.
Sources
- Writer. Adore Me customer story. A case study published by the software vendor. writer.com
- Google. Bayer: search trends and demand predictions. A case study published by Google. business.google.com
- US Copyright Office (January 2025). Copyright and Artificial Intelligence, Part 2: Copyrightability. copyright.gov
- European Union. AI Act, Regulation (EU) 2024/1689, Article 50 (transparency obligations). Text on the European Commission’s AI Act Service Desk. ai-act-service-desk.ec.europa.eu
- European Union. AI Act, Article 111 (systems already on the market), with the Commission’s summary of the December 2026 date. ai-act-service-desk.ec.europa.eu
- US Federal Trade Commission (14 August 2024). Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials. Rule: 16 CFR Part 465. ftc.gov
- Hannane, Demirci and Zhu (2025). Who Is AI Replacing? The Impact of Generative AI on Online Freelancing Platforms. Management Science 71(10). doi.org
- Goldberg and Lam (2025). Generative AI in Equilibrium: Evidence from a Creative Goods Marketplace. Marketing Science Institute Working Paper 25-129. thearf-org-unified-admin.s3.amazonaws.com
- Federal Reserve Banks (March 2026). Small Business Credit Survey: 2026 Report on Employer Firms. fedsmallbusiness.org
- Advertising Standards Authority (11 June 2026). AI and Deepfakes: Four Things Advertisers Need to Know Before They Hit ‘Run’. asa.org.uk