UNIT.City — місце, де люди працюють... КРАЩЕ! Обирай свій простір просто зараз 👉
Ігор Вишневський AI Eng
23 March 2026, 12:42
2026-03-23
How much do people who sell their videos and phone conversations to train AI earn? What are the prospects for the new "profession"?
The Guardian analyzed the market for services that pay people to sell their personal photos, videos, and phone conversations, which are then used to train artificial intelligence.
The Guardian analyzed the market for services that pay people to sell their personal photos, videos, and phone conversations, which are then used to train artificial intelligence.
For example, South African Jacobus Lowe sold a video of himself walking along the waterfront in Cape Town to the Kled AI service for $14, which is 10 times the country’s minimum wage per hour and could provide him with half a week’s worth of groceries.
The video was about a «City Navigation» challenge that Lowe found on the service, which pays creators to upload their data, such as videos and photos, to train artificial intelligence models. Over the course of a few weeks, Lowe earned $50 by uploading photos and videos of his daily life.
Another example cited by the publication in its article is Sahil Tigga, a 22-year-old student from India. He regularly earns money by giving Silencio, a company that crowdsourcing audio data for training artificial intelligence, access to his phone’s microphone. This records ambient city noise, traffic at an intersection, or the student uploads recordings of his voice.
Sahil travels the city specifically to capture sounds in unique locations, such as hotel lobbies, that are not yet documented on the Silencio map. He earns over $100 a month, enough to cover all his food expenses.
Another example is 18-year-old welding apprentice Ramelio Gill from Chicago, who made several hundred dollars selling his private phone conversations with friends and family to Neon Mobile, a conversational AI training platform that pays $0.50 per minute.
Hill had mixed feelings about selling his private phone calls to Neon Mobile. He made $200 in about 11 hours of calls, but he said the app frequently went offline and didn’t pay past-due bills. «Neon was always suspicious to me, but I kept using it to get some extra easy money for various expenses,» Hill said.
Pitfalls of a new profession
The Guardian writes that Silicon Valley’s «hunger» for high-quality data exceeds what can be collected from the internet, so a thriving industry of data marketplaces has now emerged that pay people for content.
Experts estimate that companies developing artificial intelligence will be left without fresh, high-quality content for training as early as 2026.
Therefore, content creators for artificial intelligence, who upload everything from the scenes around them to photos, videos, and audio recordings of themselves, are at the forefront of a new global data «gold rush.»
This is where apps like Kled AI and Silencio come in: marketplaces where millions of people monetize their data to feed and train AI. In addition to Kled AI, Silencio, and Neon Mobile, there are many options for «AI trainers»: for example, Luel AI, backed by the renowned startup incubator Y-Combinator, provides multilingual conversations for around $0.15 per minute.
ElevenLabs, meanwhile, provides a digital clone of anyone’s voice and allows anyone to use it for a base fee of $0.02 per minute.
In turn, AI training companies are realizing that paying people to license their data helps them avoid the risk of copyright disputes they might face if they relied entirely on content gathered from the internet.
People in developing countries often need money and have few other options for earning it. For many «AI trainers,» this work is a pragmatic response to economic inequality: in countries with high unemployment and devaluation of currencies, such earnings are often more stable and profitable than local work.
However, companies that buy such data often set traps: on some AI marketplaces, content creators grant irrevocable, royalty-free licenses that allow companies to create «derivative works,» meaning a 20-minute voice recording uploaded today could power an AI-powered customer service bot for years to come, without the original content author receiving a cent.
Selling your content can also have significant privacy implications. For example, one such content creator we spoke to was the subject of an AI video in which he was presented as a gynecologist and used to promote unproven health supplements for pregnant and postpartum women.
Чатбот від Google заявив інженеру, що живий, а той повірив. Чи загрожує нам повстання роботів
Інженер Блейк Лемуан, що працював у відділі відповідального штучного інтелекту Google і розмовляв з чатботами каже, що одна із систем штучного інтелекту компанії — мовна модель The Language Model for Dialogue Applications (LaMDA) — може мати власні почуття. У червні про це з’явилась стаття у Washington Post.