AI Ethics Intern (Evaluation, Guardrails & Responsible AI)

Description

Sony Research is dedicated to driving innovation, advancing technology and ensuring that AI is developed responsibly, ethically, and inclusively. The AI Ethics team focuses on creating frameworks, tools, and methodologies that promote trust, transparency, fairness, and accountability in AI tools and technologies.

We are seeking a motivated Research Intern to join our interdisciplinary team at the forefront of AI Ethics, Safety, and Responsible AI for a 3-month research internship. You will work on open research questions in generative AI, including data collection, guardrails, evaluation, and benchmarking to tackle fundamental ethical challenges that arise when AI is deployed at scale in a global entertainment company. This is an opportunity to conduct meaningful, publishable research in collaboration with experienced researchers and engineers across a globally diverse team. We actively support interns in co-authoring papers and submitting to top-tier venues, making this an ideal opportunity to advance your doctoral research while contributing to real-world impact.

Responsibilities

Responsibilities will depend on the project you are assigned to, and will include a combination of the following:

• Conduct innovative research in AI ethics, including but not limited to data collection, AI evaluation and harm mitigation.
• Contribute to the development of tools and frameworks to assess and mitigate AI-related risks in natural language processing and computer vision, such as bias, privacy, copyright and transparency.
• Improve the performance of open models to detect and mitigate AI harms.
• Set up technical experiments to conduct evaluations of open models.
• Contribute to research with the potential for publication.

Qualifications

• Currently pursuing a degree (PhD preferred) in Computer Science or socio-technical AI topics.
• Strong foundations in AI and machine learning techniques in computer vision or natural language processing.
• Familiarity with generative AI evaluation, benchmarking, red-teaming or guardrail development.
• Experience using Python and deep learning libraries (e.g., PyTorch, HuggingFace Transformers) to develop, fine-tune, or evaluate generative AI models.
• Strong interest in AI ethics and responsible AI development.
• Proven research, analytical, and problem-solving skills.
• Self-motivated and capable of proposing and implementing innovative ideas.
• Excellent written and verbal communication skills in English.
• Preferred: Experience with research communities, including having published papers at conferences/journals

Preference will be given to candidates based on-site in Zurich; however, remote applicants located in the EU, UK, or the United States may also be considered. Applicants must ensure they are legally authorized to work in the country from which they apply, as visa sponsorship is not provided.

Start date: 15 June 2026 or as agreed
Duration: 3 months