AI Safety Experts — English & Bengali | $20-$22/hr Remote
Overview
This role brings fluent English and Bengali speakers into a remote AI red team. You'll probe conversational AI systems with adversarial inputs to uncover biases, misinformation, and manipulation risks. All work is text-based, and you'll produce structured red team data that improves AI safety. You can opt out of higher-sensitivity topics, and clear guidelines plus wellness resources are provided.
What You'll Do4
- 1Pressure-test conversational AI models using jailbreak techniques, prompt injection, and bias exploitation to surface failure modes.
- 2Create high-quality human data by labeling model failures, categorizing vulnerability types, and flagging systemic risks.
- 3Apply established taxonomies, benchmarks, and playbooks to keep your adversarial testing structured and repeatable.
- 4Document findings in clear reports and prepare attack-case datasets that customers can act on.
Requirements5
- 1Native-level fluency in English and Bengali for written and spoken communication.
- 2Hands-on experience with adversarial AI testing, cybersecurity, or socio-technical risk probing.
- 3A systematic mindset: you use frameworks, benchmarks, or playbooks rather than random testing.
- 4Strong ability to explain technical risks to both technical and non-technical audiences.
- 5Adaptability to work across diverse projects and customer environments.
Who Should Apply
You have a natural curiosity for how systems fail and a structured approach to probing them. Your background may include adversarial machine learning, penetration testing, abuse analysis, or creative fields like psychology and writing. You're comfortable handling sensitive content and following established guidelines. You want a remote, hourly role where your adversarial thinking shapes safer AI.
Salary Insight
Pay is $20.00 - $22.00 per hour for this remote, hourly engagement.
Location
Required Skills
Application Tip
In your application, describe one instance where you successfully jailbroke or exposed a flaw in an AI system. Explain the approach you took and what you learned.
See NearSkill jobs more often in your search
How your application is processed
1Application received
Your resume and details are logged the moment you apply.
2ATS + eligibility screening
We check your profile against the role’s skills, seniority, and requirements.
3Employer sees qualified profiles only
Only candidates who clear screening move forward.
Similar open positions
Explore active roles that match your skills and interests.
Mercor
VerifiedAI Safety Experts — English & Odia
We're assembling a red team of human experts to stress-test conversational AI models for safety vulnerabilities. This remote, text-based position requires native fluency in English and Odia. You'll attack AI systems with adversarial inputs, uncover weaknesses, and generate the high-quality data needed to make AI safer. The work covers sensitive topics like bias and misinformation, with clear guidelines and wellness support available.
Mercor
VerifiedAI Safety Experts — English & Punjabi
Join a remote red team that stress-tests AI models for safety vulnerabilities. You will probe conversational AI with adversarial inputs, uncover weak spots like jailbreaks or bias, and turn those findings into data that makes AI safer for customers. Native-level fluency in English and Punjabi is a must. The work is fully text-based and paid hourly, with optional exposure to sensitive topics supported by clear guidelines.
Mercor
VerifiedAI Safety Experts — English & Malay
This remote role asks bilingual experts in English and Malay to stress-test AI systems by simulating attacks and probing for weak points. You'll generate high-quality red team data that helps customers build safer, more trustworthy models. All work is text-based, and sensitive topics are clearly disclosed with wellness support.
Mercor
VerifiedAI Safety Experts — English & Assamese
Mercor is assembling a remote red team of human experts to attack AI models with adversarial inputs. This role focuses on conversational AI, using native English and Assamese to probe for jailbreaks, prompt injections, and bias exploitation. You'll generate red team data and vulnerability reports that make AI systems safer for customers. The work is entirely text-based, with optional participation in higher-sensitivity projects supported by clear guidelines and wellness resources. Hourly compensation is $20-$22.
Mercor
VerifiedAI Safety Experts — English & Norwegian
This remote role invites fluent English and Norwegian speakers to join an elite red team probing AI models for vulnerabilities. You'll simulate adversarial attacks, document exploits, and generate high-quality human data that directly strengthens AI safety. The work is text-based and focuses on sensitive topics like bias and misinformation, with optional high-sensitivity projects supported by clear guidelines.
Mercor
VerifiedAI Safety Experts — English & Swedish
This role involves stress-testing AI systems by simulating adversarial attacks in English and Swedish. You'll generate critical safety data by probing models for vulnerabilities like bias, misinformation, and security flaws. As part of a human red team, you'll help ensure AI behaves safely before it reaches users.