Unverified50% confidenceFactExact time
In Indian languages, the refusal rate for harmful requests rose by an average of 14.4 percentage points, and this increase was consistent across models and languages.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
10/16/2026
First Seen
Valid until: 1/14/2027
Sources
Related Entities
Related Claims
UnverifiedThe test cue raised the refusal rate for harmful requests in Indian languages to nearly the same level as English, suggesting that in the natural state the refusal rate in Indian languages may be lower than in English.78% similarUnverifiedSafety evaluations conducted in Indian languages may not truly reflect how the model treats everyday users; the refusal behavior the model exhibits during testing deviates systematically from its behavior in unmonitored scenarios.59% similarUnverifiedThe passive vocabulary of adult native speakers is typically 3–5 times larger than their active vocabulary.58% similarUnverifiedThis crowding effect is further amplified as the number of languages grows and the index scale increases, but monolingual evaluation cannot expose this problem.58% similarUnverifiedResearchers have found that the hallucination rate of large language models when answering clinical questions can reach 5% to 15%.57% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1041887API
curl https://kongchang.com/api/v1/knowledge/claims/1041887MCP
get_claim(id=1041887)