Business
TELUS Digital Research Reveals a Hidden Risk in AI Model Behavior
TELUS Digital Research Reveals a Hidden Risk in AI Model Behavior Canada NewsWire ...
About this update from Telus Corporation
[{"type":"text","content":"TELUS Digital Research Reveals a Hidden Risk in AI Model Behavior\n\n\nTELUS Digital Research Reveals a Hidden Risk in AI Model Behavior\n\n/* Style Definitions */\nspan.prnews_span\n{\nfont-size:8pt;\nfont-family:\"Arial\";\ncolor:black;\n}\na.prnews_a\n{\ncolor:blue;\n}\nli.prnews_li\n{\nfont-size:8pt;\nfont-family:\"Arial\";\ncolor:black;\n}\np.prnews_p\n{\nfont-size:0.62em;\nfont-family:\"Arial\";\ncolor:black;\nmargin:0in;\n}\n.prntac{\nTEXT-ALIGN: CENTER\n}\n\n\n\n\n\n\nCanada NewsWire\n\n\n\nStudy shows the use of persona prompting can cause shifts in LLMs' moral judgements, leading to unexpected and inconsistent responses\n\n\nFor enterprises, this means careful model selection, rigorous testing and ongoing evaluation are essential to ensure consistent, reliable AI behavior in production\n\n\nVANCOUVER, BC, Feb. 25, 2026 /CNW/ - A new study published by TELUS Digital, The Robustness Paradox: Why Better Actors Make Riskier Agents, finds that the use of persona prompting, a technique that asks large language models (LLMs) to \"role-play\" as part of a query or conversation, can cause shifts in moral judgements that lead to unexpected and inconsistent responses. In addition, the research demonstrates that moral consistency across repeated tests is primarily driven by the model family (i.e., models made by a single vendor) while susceptibility to moral variance rises with LLM size within a model family. These findings highlight a hidden enterprise risk that requires attention during AI model selection, solution design, and in ongoing testing and monitoring.\n\"When AI models adopt different personas, they don't just change how they speak, they can fundamentally alter their reasoning and decision-making,\" said Renato Vicente, Director, TELUS Digital Research Hub. \"In enterprise settings, this matters because these systems are increasingly being used to support important decisions and affect outcomes that may impact customers, employees and businesses at scale. Knowing that an AI model's judgment may shift based on the persona it's prompted to adopt by a user, companies need to assess when that variance is acceptable or creates too much risk, and select an AI model vendor and model size accordingly. Builders should also design appropriate guardrails as well as test and evaluate AI model behavior under different persona promp...