Director, AI Security

Johnson & Johnson Innovative Medicine•New Brunswick, NJ

1d•$154,000 - $265,650

About The Position

The Director, AI Security is a cybersecurity leader accountable for establishing and scaling enterprise AI security capabilities across Johnson & Johnson. The role covers security for enterprise AI platforms, commercial SaaS AI services, AI embedded in third-party products, open-source models and components, internally developed AI solutions, and agent ecosystems. Its aim is to enable secure AI adoption while protecting sensitive information and critical operations in a regulated healthcare environment, including GxP/validated systems, HIPAA, and the EU AI Act. The position leads a team responsible for identifying and mitigating AI-related risks, including prompt injection, sensitive data exposure, excessive agent privileges, tool misuse, model and data poisoning, and compromised AI supply chains. It validates control effectiveness through adversarial testing and works with IAM, Vulnerability Management, Application Security, Cyber Defense, and GSO on identity security, remediation, threat detection, and response. The role partners with AI Governance, Privacy, Security and Technology Architecture, and Risk Management to embed security controls through enterprise standards, onboarding, and operational guardrails; governance and architecture decisions remain with those partner teams. The Director also operationalizes frontier and other advanced models to identify and prioritize cybersecurity exposures, accelerate investigations, and expedite remediation with safeguards and human validation. Success is measured by expanded security coverage, effective protection of AI systems and models, faster risk identification and remediation, and demonstrable reduction in enterprise cybersecurity risk.

Requirements

Bachelor's degree in Information Security, Computer Science, Engineering, Information Systems, or a related discipline, or equivalent experience.
12+ years of progressive experience in cybersecurity, application security, security engineering, or related disciplines, including leadership of enterprise security capabilities; this experience requirement is overall cybersecurity experience, not AI-specific experience.
Recent hands-on experience securing AI-enabled systems.
Demonstrated experience implementing, operating, and scaling AI security capabilities across end-user generative AI services, AI applications, agents, and models, with measurable improvements in coverage and risk reduction.
Working understanding of AI and machine learning lifecycles, large language models, retrieval-augmented generation (RAG), model APIs, and agent tool integrations, including their security implications across development, deployment, and inference.
Practical expertise in AI discovery, identity and access enforcement, least-privilege agent permissions, secrets protection, data loss prevention, and runtime safeguards.
Demonstrated ability to assess and mitigate prompt injection, sensitive data exposure, excessive agency, tool misuse, insecure output handling, model theft, adversarial manipulation, data and model poisoning, and AI supply-chain compromise.
Experience in cloud and platform security, including securing the cloud, data, and identity platforms used by AI workloads and services.
Experience integrating AI security telemetry with security operations and incident response in partnership with Cyber Defense, including detection engineering, investigation, containment, and remediation of threats affecting AI systems and models.
Proven ability to lead technical security teams, manage vendors and budgets, oversee concurrent capability deployments, and partner across engineering, security operations, and business stakeholders in a global, highly matrixed enterprise.
Strong executive stakeholder management, technical judgment, and communication skills, including the ability to translate emerging AI threats into prioritized defenses and communicate business impact, control effectiveness, security coverage, and remediation outcomes.

Nice To Haves

Advanced degree (MBA, MS, or equivalent).
Experience securing production multi-agent systems and Model Context Protocol (MCP) integrations, including delegated authorization, isolated tool execution, inter-agent trust boundaries, and defenses against memory poisoning and cascading compromise.
Deep technical experience evaluating open-weight and fine-tuned models for backdoors, poisoning, model extraction, and training-data leakage, and building automated adversarial evaluation suites informed by NIST adversarial machine learning guidance, MITRE ATLAS, and OWASP AI security guidance.
Experience embedding AI security testing and controls into software delivery and MLOps pipelines, including model and dependency assessment, adversarial testing, AI red teaming, remediation validation, and repeatable evaluation of models, applications, and agent workflows.
Advanced machine learning engineering or frontier model research experience, including model development, fine-tuning, and evaluation, and adversarial AI research or specialized red-team experience.
Working knowledge of the NIST AI Risk Management Framework and Generative AI Profile, OWASP guidance for LLM and agentic application security, and MITRE ATLAS, with the ability to translate threats and recommended practices into operational controls and security tests.
Practical experience applying frontier AI or other advanced models to cybersecurity use cases, including exposure identification, risk prioritization, investigation, and remediation automation; experience moving frontier-model-powered vulnerability discovery and remediation from pilot to sustained production use, with validation of generated fixes, bounded execution, approval checkpoints, rollback, and quantified improvements in remediation speed and quality.
Experience scaling AI security across cloud-hosted and self-hosted models in regulated healthcare, pharmaceutical, or MedTech environments, including protection of research, clinical, or manufacturing data and validation of controls without disrupting critical workflows.
Demonstrated contributions to the AI security field through responsibly disclosed vulnerabilities, published applied research, open-source security or evaluation tools, or substantive contributions to OWASP, MITRE ATLAS, or comparable industry initiatives, with evidence of translating findings into deployed defenses.
Experience with skill security and multi-turn agentic (loop prompt) security.
Relevant certifications such as CISSP, CISM, or equivalent credentials; AI security certifications such as GIAC AI Platform Security (GAIPS), GIAC AI Security Automation Engineer (GASAE), or ISACA Advanced in AI Security Management (AAISM) are particularly desirable.

Responsibilities

Establish and scale enterprise AI security capabilities covering LLM models and harnesses, lifecycle management, agents and non-human identity, end-user AI usage, AI supply chains, runtime protection, lifecycle assurance, observability and monitoring, and AI application security.
Establish AI discovery capabilities to identify enterprise use of generative AI, AI applications, agents, and models and prioritize gaps in security coverage.
Establish end-user AI usage management, including access enforcement and data protection controls, to prevent unauthorized access and exposure of sensitive information.
Own AI agent identity and authorization security, including non-human identities, credentials, service accounts, and least-privilege permissions, in partnership with the IAM team.
Partner with stakeholder teams to integrate security testing and protective controls throughout AI system development, deployment, and operation.
Establish AI onboarding and assurance, including security evaluations of new capabilities before production, model assessments and ongoing management, agent security reviews, AI red-team guidance, supply-chain assurance, and runtime monitoring requirements.
Define security approval criteria for production deployment of AI models, agents, and applications, including emerging and frontier model capabilities.
Implement model and AI supply-chain security controls to detect and mitigate model or data poisoning, compromised components, and exposures affecting deployed models.
Ensure continuous adversarial testing of models, AI applications, and agent workflows to identify prompt injection and other high-impact threats, validate defenses, and verify remediation.
Integrate AI security telemetry and detections into security operations and partner with Cyber Defense on AI-specific detection, triage, and incident response playbooks.
Deploy frontier AI and other advanced models to identify and prioritize cybersecurity exposures, accelerate investigations with Cyber Defense and GSO, and expedite remediation with safeguards and human validation.
Partner with AI Governance, Privacy, Security and Technology Architecture, and Risk Management to enable security controls through enterprise AI standards, onboarding processes, and operational guardrails.
Own AI security vendor relationships and budget, including vendor selection, performance management, spend, vendor strategy, and capability rationalization.
Partner with Vulnerability Management, Application Security, and business teams to drive timely remediation of AI-related exposures.
Be accountable for enterprise AI security standards adoption, measurement of control effectiveness, executive risk communication, and the AI security roadmap and investment planning in partnership with Security Architecture.
Lead technical security teams, set clear ownership and escalation paths, decision rights, quality standards, and delivery expectations, and foster mentoring, knowledge sharing, collaboration, and continuous improvement.

Benefits

Consolidated retirement plan (pension) and savings plan (401(k)), subject to plan terms.
Eligibility to participate in the long-term incentive program.
Vacation: 120 hours per calendar year, subject to applicable policy and date of hire.
Sick time: 40 hours per calendar year; 48 hours for Colorado residents and 56 hours for Washington residents, subject to applicable policy and date of hire.
Holiday pay, including floating holidays: 13 days per calendar year.
Work, Personal and Family Time: up to 40 hours per calendar year.
Parental leave: 480 hours within one year of the birth, adoption, or foster care of a child.
Bereavement leave: 240 hours for an immediate family member and 40 hours for an extended family member per calendar year.
Caregiver leave: 80 hours in a 52-week rolling period.
Volunteer leave: 32 hours per calendar year.
Military spouse time off: 80 hours per calendar year.