AI Safety, Responsible AI & Cybersecurity Specialist (AI Evaluation)
The middle of its range, $115/hr, is above the $90/hr median of the Red-Teaming & AI Safety roles open now, whose typical pay runs $50–$150/hr. How that pay moves week by week is in our research on Red-Teaming & AI Safety work.
Half of all Red-Teaming & AI Safety roles come down within 25.5 days of being listed, and 19.7% are still up after four weeks. This one was posted today.
About the role
1. Role Overview
Join a leading AI lab’s cutting-edge Generative AI team and contribute to the development of next-generation foundational AI models. We are seeking experienced AI Safety, Responsible AI, and Cybersecurity Specialists with practical expertise in application security, offensive security, threat analysis, red teaming, or trust and safety.
In this role, you will bring an adversarial, security-focused, and safety-first perspective to AI model development and evaluation. You will help identify vulnerabilities, assess potentially harmful model behaviour, design realistic adversarial scenarios, and establish evaluation standards to improve the security, reliability, and responsible use of AI systems.
The ideal candidate combines strong cybersecurity or trust and safety expertise with analytical thinking and the ability to assess complex technical and ethical risks in AI-generated outputs.
This is a W-2 employment opportunity, with the opportunity to work as part of a leading AI lab’s extended workforce.
2. Key Responsibilities
- Collaborate with research and engineering teams to identify knowledge gaps and improve AI model performance across cybersecurity, Responsible AI, application and infrastructure security, red teaming, and abuse prevention.
- Design challenging, domain-specific tasks based on real-world security and AI safety scenarios, developing accurate and well-structured solutions.
- Evaluate AI-generated security and Responsible AI outputs for technical accuracy, safety, policy compliance, and potential misuse.
- Create adversarial prompts, security challenges, and red teaming scenarios to test AI systems for vulnerabilities, unsafe responses, harmful behaviour, and policy violations.
- Assess AI systems against potential threats, including prompt injection, data leakage, insecure code generation, harmful instructions, and other relevant attack patterns.
- Develop evaluation frameworks, scoring rubrics, and quality standards to assess the safety, reliability, accuracy, and potential harm of AI-generated responses, particularly on security-sensitive topics.
- Identify weaknesses in AI model responses and provide clear, structured, and actionable technical feedback to support improvements.
- Apply relevant cybersecurity and trust and safety methodologies to evaluate model behaviour across different threat scenarios and sensitive use cases.
- Work closely with other subject matter experts to maintain consistency, accuracy, and quality across AI training datasets.
3. Core Qualifications & Experience
- 4+ years of professional experience in Cybersecurity, Application Security, Red Teaming, Penetration Testing, Threat Intelligence, Incident Response, or Trust and Safety, preferably within a recognised technology company, cybersecurity organisation, or established enterprise.
- Hands-on expertise in at least one of the following areas:
- Application Security: Secure code reviews, vulnerability assessments, application security testing, security architecture, and secure software development practices.
- Offensive Security & Penetration Testing: Vulnerability discovery, exploitation techniques, penetration testing methodologies, adversarial testing, and security assessments.
- Threat Intelligence & Incident Response: Threat analysis, attack investigation, security monitoring, incident handling, threat modelling, and risk assessment.
- Trust & Safety: Content moderation, platform abuse prevention, policy enforcement, harmful content identification, safety evaluations, and risk mitigation.
- Strong understanding of cybersecurity principles, common attack vectors, vulnerability management, and security best practices.
- Practical experience identifying and evaluating security vulnerabilities across applications, systems, infrastructure, or AI-powered environments.
- Familiarity with AI safety and Responsible AI concepts, including model security, adversarial robustness, harmful-output prevention, privacy risks, bias, and policy compliance, is highly desirable.
- Ability to design adversarial scenarios and evaluate AI-generated responses for security weaknesses, unsafe behaviour, and potential misuse.
- Relevant industry certifications such as OSCP, OSWE, GPEN, or CISSP will be an advantage.
- Published Common Vulnerabilities and Exposures (CVEs), recognised bug bounty findings, security research, technical publications, or meaningful contributions to cybersecurity projects are strong advantages.
- Demonstrable career progression, with increasing responsibility in security engineering, offensive security, threat analysis, or trust and safety functions.
- Strong analytical and problem-solving abilities, with the capacity to investigate complex security issues and communicate findings effectively.
- Ability to commit reliably to at least 20 hours per week during weekdays .
- Excellent written communication skills, with the ability to explain complex technical concepts, security findings, risk assessments, and evaluation decisions clearly.
Where you can work
This role is 100% remote, for applicants located in United States.
From Weekday's job posting, checked 17 Nov 2026.
The application
Just your details and CV.
From Weekday's apply page, checked 17 Nov 2026.
About Weekday
A recruiting company that staffs AI training companies' expert projects, most of them open in the US and paid by the approved hour or task.
How it pays: 14 of its 90 AI-training postings say when they pay, all weekly (12 of those posted on 12 Nov 2026 through Stripe or Wise); the rest do not say. Its 2025 postings for an AI lab said independent contractors were paid weekly (daily on two) through Stripe Connect (its Workable board, read 15 Nov 2026).
When it pays: Most postings name no schedule. How often others pay
Interview: on some roles or projects (Per-posting screening (AI interview, timed tests or portfolio review) and a manager's review). Platforms with no interview
Similar open roles
Some links on AITraining.jobs are referral links. If you sign up through one, AITraining.jobs may earn a commission at no extra cost to you. It never changes what a platform pays you. A role at a platform that pays us a referral fee weighs more in this order.

Prompt & Response Evaluator (opens the job details)
micro1 · AI Evaluation · Remote· 4d agoDetails
$40–$70/hr
Document Annotation Specialist (PDF) (opens the job details)
SuperAnnotate (SME) · Data Annotation · Remote· 6d agoDetails
up to $10/hr
Swift Engineer (iOS Code Review) (opens the job details)
OpenTrain AI · Coding · Remote· 4mo agoDetails
up to $55/hr
Full-Stack Software Engineer - Testing & Evaluation (opens the job details)
AfterQuery · AI Evaluation · Remote· 11mo agoDetails
$75/hr
Generalist AI Trainer (opens the job details)
RemoExperts · AI Evaluation · Remote· 1y+ agoDetails
$30–$65/hr
Want companies to ask for you?
One profile for every AI training platform. Companies ask us for people with a specific language, country or credential: add yours once, and when a role fits, we introduce you, only with your yes. Free.
Get new Weekday roles by email
We check the Weekday job board every hour and email you its newest roles each week, plus similar paid roles elsewhere. Free, unsubscribe anytime.