Responsible AI & Governance Technical Specialist

Written by the T&S Network from Sony's posting.

Responsible AI and governance work for generative and agentic AI products at a game and entertainment company, sitting in a cross-functional program that embeds ethical and safety review into product development. The role advises teams on AI risk across engineering, product, information security, and legal, aiming to align AI initiatives with regulatory requirements and internal strategy.

• AI misuse and jailbreaks

What you'd do

• Advise teams on agent designs, tool connections, and data access in generative and agentic systems as a technical RAI adviser
• Assess autonomy, human oversight, and guardrails for agentic AI and recommend mitigations
• Carry out ethical and safety risk reviews that examine bias, equity, reliability, openness, and broader social effects
• Experiment hands-on with agents, MCP servers, APIs, and evaluation frameworks to turn findings into governance guidance
• Coordinate AI governance reviews with Information Security to align them with security controls
• Sit on domain boards as a technical specialist and work with Product and Engineering on risk remediation
• Contribute best practices, AI literacy materials, and case studies for company-wide responsible AI development

What it asks for

• Three-plus years in AI governance, responsible AI, AI safety, AI policy, or adjacent work, including evaluating AI systems
• Five or more years building multi-team programs, from strategy to day-to-day rollout, alongside product, engineering, legal, security, and operations
• Strong technical fluency in LLMs, agentic architectures, tool use, APIs, cloud services, and automation
• Hands-on experience evaluating agentic AI systems
• Familiarity with MCP or comparable agent and tool integration protocols (preferred)
• Experience with AI red teaming or adversarial testing of LLM and agentic systems
• Knowledge of agentic and sociotechnical risks such as excessive agency, permissions, and emergent behaviors

Worth knowing

• Subject to a background check at offer stage, which may include criminal background checks for some roles

Terms in this role

Red teaming.

Testing a system, often an AI model, by deliberately trying to make it fail or misbehave, so problems are found and fixed first.

Our best summary of the posting, and it can be wrong. The employer's posting is what counts: read it in full on Sonynorth_east before you apply. If we got something wrong, report this listing.