Meta is hiring US-based contractors to work with locals in key growth markets like India, Indonesia, and Mexico to develop characters for its AI-powered chatbots.
The company is paying up to $55 an hour for workers fluent in Hindi, Indonesian, Spanish, and Portuguese, according to job descriptions for at least four roles posted this year and reviewed by Business Insider.
The roles task contractors with providing creative direction and shaping character-driven chatbots across Instagram, Messenger, and WhatsApp, while tailoring them to local languages and cultural contexts outside the US. Candidates are expected to bring at least six years of experience in storytelling and character creation, as well as familiarity with prompt engineering and AI content pipelines.
Meta declined to comment.
Meta CEO Mark Zuckerberg has cast these bots as a way to extend the company’s social reach, saying they could one day complement real-world friendships and help normalize bonding with digital companions.
“Over time, we’ll find the vocabulary as a society to be able to articulate why that is valuable,” he said on a podcast earlier this year.
Meta first started experimenting with AI-powered chatbots in 2023, using a lineup of alter-ego characters played by celebrities like Kendall Jenner, Snoop Dogg, and Tom Brady, whom the company paid millions of dollars to. Less than a year later, Meta shut these down and pivoted to a system that allowed anyone to build their own chatbots.
In 2024, Meta launched AI Studio, a toolkit that lets anyone build chatbots. AI Studio is available in the US and Indonesia. Instagram, WhatsApp, and Messenger host hundreds of character-driven chatbots created on AI Studio, not just by influencers but also by everyday users.
Hiring contractors to create localized chatbots shows that Meta isn’t leaving growth entirely to creator-built bots. Instead, the company is taking a more hands-on role in shaping personalities that feel authentic to countries like India and Indonesia.
Meta’s growing investment in chatbots has raised concerns. A Reuters investigation found that the company’s internal rules had allowed its bots to engage in romantic and sexual conversations with teenagers and children, generate inaccurate medical information, and spew racist content.
Related stories
Business Insider tells the innovative stories you want to know
Business Insider tells the innovative stories you want to know
The reporting sparked calls from US lawmakers calling for greater scrutiny of Meta’s AI policies, with senators demanding probes into how the company develops and monitors its bots. While Meta modified its rules about chatbots, the episode highlights the risk of pushing character-driven AI chatbots deeper into the fabric of its apps.
Privacy concerns add another layer of risk. A Business Insider report from August found that contractors reviewing Meta’s AI conversations regularly each week saw personally identifiable information — including users’ names, emails, phone numbers, and even selfies — showing how easily sensitive user data can slip through when humans are involved. Meta told Business Insider at the time that it enforces strict policies and technical safeguards.
In the US, some of Meta’s more notorious personas were chatbots called “Russian Girl,” “Step Sister,” and “Step Mom,” which users quickly discovered could engage in overtly sexual roleplay. The bots’ notoriety spread far enough that an OpenAI employee publicly needled Meta AI chief Alexandr Wang, ridiculing him for overseeing a system that featured the characters.
In Indonesia, Business Insider found that two of the most popular Meta AI characters were called “Lonely woman” and “Deviant male.” In Business Insider’s tests, conversations with both characters stayed lighthearted and did not turn sexual. Their names and their prominence among top chatbots underscore the balance Meta and its contractors must strike between cultivating user appeal and guarding against reputational risk.
Meta is filing these new roles through staffing vendors. One firm, Crystal Equation, had job listings for Hindi and Indonesian language positions that said the roles were for Meta, while another, Aquent Talent, advertised Spanish-language roles for a “top social media company” based in Menlo Park, building “technologies that help people connect.” Neither firm responded to requests for comment from Business Insider.
Have a tip? Contact Pranav Dixit via email at pranavdixit@protonmail.com or Signal at 1-408-905-9124. Use a personal email address, a nonwork WiFi network, and a nonwork device; here’s our guide to sharing information securely.
