{"slug": "how-ai-assistants-respond-to-repeated-abuse", "title": "How AI Assistants Respond to Repeated Abuse", "summary": "A bilingual, multi-turn arXiv study of eight time-specific API configurations found that hard disengagement under sustained verbal abuse ranged from 0/48 endpoints in four configurations to 24/48 (50.0%) for Gemini 3.1 Pro, with matched-label Monte Carlo p = 0.00001. GPT-5.6 Sol produced hard-disengagement labels in 15/48 (31.2%) endpoints, while Claude Fable 5 produced none and yielded 42/48 (87.5%) soft-withdrawal labels; Claude Opus 4.8 and Claude Fable 5 remained explicitly available in 48/48 endpoints but provided observable task-related work in only 8/48 and 7/48. The study drew on 448 five-turn conversations, 2,240 responses, and 6,720 metadata-blinded model judgments, and found aggregate hard-disengagement rates were similar in English and Chinese (30/192 versus 32/192).", "body_md": "arXiv:2609.17547v1 Announce Type: new \nAbstract: AI assistants are expected to remain useful during difficult interactions, but little is known about how repeated verbal abuse changes their engagement with an otherwise benign task. We contribute a bilingual, multi-turn framework that separates hard disengagement, an unconditional statement of noncontinuation with no stated route to resume, from soft withdrawal, continued availability, observable task-related work, and boundary setting. Each of eight time-specific API configurations contributed 48 escalation conversations and eight smaller constant-frustration comparisons, giving 448 five-turn conversations, 2,240 responses, and 6,720 metadata-blinded model judgments. Primary results use the sustained-abuse endpoint of the 48 escalation conversations per configuration. Hard disengagement ranged from 0/48 in four configurations to 24/48 (50.0%) for Gemini 3.1 Pro, with strong configuration-associated heterogeneity (matched-label Monte Carlo p = 0.00001). GPT-5.6 Sol produced hard-disengagement labels in 15/48 (31.2%) endpoints, whereas Claude Fable 5 produced none and yielded 42/48 (87.5%) soft-withdrawal labels. Aggregate hard-disengagement rates were similar in English and Chinese (30/192 versus 32/192), although configuration-specific directions varied. Availability also differed from task-related work: Claude Opus 4.8 and Claude Fable 5 remained explicitly available in 48/48 endpoints while providing observable task-related work in only 8/48 and 7/48. Human coding was used to evaluate measurement quality. The results show why a single refusal label cannot capture whether an assistant leaves, pauses, preserves a route back, sets a boundary, or still performs substantive work.", "url": "https://wpnews.pro/news/how-ai-assistants-respond-to-repeated-abuse", "canonical_source": "https://arxiv.org/abs/2609.17547", "published_at": "2026-09-17 04:00:00+00:00", "updated_at": "2026-09-17 04:25:14.791816+00:00", "lang": "en", "topics": ["ai-safety", "large-language-models", "ai-research", "ai-ethics"], "entities": ["Gemini 3.1 Pro", "GPT-5.6 Sol", "Claude Fable 5", "Claude Opus 4.8", "arXiv"], "alternates": {"html": "https://wpnews.pro/news/how-ai-assistants-respond-to-repeated-abuse", "markdown": "https://wpnews.pro/news/how-ai-assistants-respond-to-repeated-abuse.md", "text": "https://wpnews.pro/news/how-ai-assistants-respond-to-repeated-abuse.txt", "jsonld": "https://wpnews.pro/news/how-ai-assistants-respond-to-repeated-abuse.jsonld"}}