AI Safety & Policy Associate - Intermediate (AI Community) at TELUS

About the job

• At TELUS Digital, we believe that safety is a continuous process, not a one-time check. As an AI Safety & Policy Associate, you are the primary guardian of our AI training datasets. You will join our AI Community as a specialist dedicated to identifying, labeling, and mitigating risks at the data level. While our Strategists design the rules, you are the one who applies them to thousands of complex scenarios to ensure the model learns what is right, what is wrong, and what is a "gray area." At TELUS, your daily work directly prevents harmful AI behaviors before they ever reach a user.

What You’ll Do (Key Responsibilities):

• Safety Annotation: Evaluate AI-generated responses against strict safety rubrics, flagging content that contains bias, misinformation, or prohibited topics.
• Harmful Content Classification: Categorize various types of "harm" (e.g., hate speech, self-harm, or illegal acts) to help the model understand specific boundary lines.
• Fact-Checking & Grounding: Verify the factual accuracy of model claims using reliable external sources to prevent "hallucinations."
• Guideline Application: Apply complex policy documents to "edge cases"—scenarios where the answer isn't clearly right or wrong—to ensure consistent model alignment.
• Feedback Loop Participation: Report recurring model weaknesses to the Governance Leads to help refine and update our global safety policies.

Qualification path

What You Bring (Mandatory Qualifications):

• Education: Minimum of a Bachelor’s Degree in any field. (Preferred: Humanities, Social Sciences, Communications, or Law).
• Native Language: Native-level proficiency in your primary language is mandatory. You must understand the cultural "slang" and subtle context that signifies harm in your region.
• English Proficiency: Minimum B1 (Intermediate) level English to follow community guidelines and log safety reports.
• Detail Orientation: An exceptional ability to spot small errors and remain focused during repetitive, high-stakes analysis.