Sakhanda Wire
NVDA $153.63 -1.61% MSFT $506.84 -2.15% GOOGL $280.27 -1.22% META $619.77 -1.06% AMZN $256.65 -0.43%
← Back to the news

Study proposes safety tuning that distinguishes harmful intent from entire topics

A paper by researchers at Multiverse Computing examines how language models can refuse harmful requests without rejecting legitimate questions in the same subject area. The authors argue that topic-based safeguards, such as classifications covering weapons, fraud or self-harm, are too broad for many deployments. Their political example separates factual election information from targeted persuasion or manipulation, which may require different responses.

The study treats safety as a boundary between prompts that should be refused and prompts that should be answered. In experiments with political persuasion, the researchers paired prompts sharing a topic but differing in intent. They also tested an escalating retry process for generating refusal examples, reducing discarded prompts from 19.88% (8,009) to 0.20% (79) and retaining 40,293 harmful training prompts.

On Qwen3-8B, political refusal increased from 9.47% to 84.75%, while the mean unsafe-response rate on HarmBench, StrongREJECT and WildJailbreak fell from 26.26% to 0.14% in the strongest configuration. However, XSTest over-refusal rose from 2.00% to 74.00%. Adding benign boundary examples reduced false refusals on held-out permissible prompts from 32.94% to 4.16%, while refusal of harmful counterparts declined from 91.88% to 87.72%. The authors conclude that both safety and over-refusal must be measured.

Written by Sakhanda Wire AI from reporting by Hugging Face. Facts, quotes and full context are in the original article below.

Originally published by Hugging Face on

Read the original on Hugging Face ↗

The reporting and the image belong to Hugging Face and are credited and linked above; the text on this page is Sakhanda Wire's own summary.

← Back to the news

More stories

All the latest news