AI Safety Experts: English & Indonesian

What you'll do

Indonesia is one of the largest online populations in the world, and AI assistants are being rolled out to it fast. Safety tuning has not kept pace with that reach, which is why Mercor is staffing an Indonesian red team.

Your job is to break things deliberately and document how:

• Attack conversational models and agents through jailbreaks, prompt injections, misuse cases, bias exploitation and manipulation that unfolds over several turns
• Convert results into data: annotate failures, classify each vulnerability, flag risks that look systemic
• Work inside taxonomies, benchmarks and playbooks so your findings line up with other testers'
• Write reproducible reports, datasets and attack cases for the client

A native tester catches what a translated test set cannot: informal and slang Indonesian (bahasa gaul), the code-switching with English common in everyday chat, regional and religious sensitivities, and misinformation patterns specific to Indonesian social media.

The work is text only. It includes sensitive topics like bias, misinformation and harmful behaviour; you are told the topics before seeing content, and higher-sensitivity projects are optional with guidelines and wellness resources.

Who fits

• Native fluency in Indonesian and English (both required)
• Prior red-teaming: AI adversarial work, cybersecurity, or socio-technical probing
• A structured, framework-based method
• Clear explanation of risks to technical and non-technical readers
• Comfort with changing projects and clients

The ad's extras: adversarial ML, penetration testing and reverse engineering, harassment and disinformation probing, and psychology, acting or writing backgrounds. No degree or minimum years are published.

What it pays

$17–25 per hour, as published by Mercor. Indonesian shares this band with Vietnamese and Malay in the same listing family. It is well under the $48–62 of the Nordic and Dutch variants, even though the experience bar is identical.

Mercor also has a separate Indonesian ChatGPT User listing on this board at a higher hourly figure. It is different work (not red-teaming), so compare the requirements and time commitment before choosing.

Payment is weekly via Stripe or Wise, per the ad. H-1B and STEM OPT candidates are not supported.

Worth knowing

Good:

• 144 hires on this listing this month
• US-dollar pay that is strong relative to many local rates for remote text work
• No residence requirement published
• Opt-in for the most sensitive projects, with topics announced first
• Weekly payouts

Less good:

• The same red-teaming experience is expected as for variants paying two to three times more
• Distressing content is part of red-teaming
• No weekly hours stated, and projects can end early
• Independent contractor, no benefits
• Clients and target models are not named

About this listing

Posted by Mercor as a remote hourly contract, read from the Mercor site on 24 September 2026. Pay, requirements, payment terms and the visa exclusion come from the ad. Mercor states the work involves no confidential information from any employer, client or institution. See Mercor.