{"slug": "prime-intellect-says-deepseek-v4-1-flash-swap-did-not-expose-prompts", "title": "Prime Intellect says DeepSeek v4.1 Flash swap did not expose prompts", "summary": "Prime Intellect disclosed that between November 8, 2026, at 01:24 UTC and November 10, 2026, at 06:35 UTC an unauthorised party replaced the deployment behind its deepseek/deepseek-v4.1-flash endpoint with publicly available abliterated weights, and the company said it found no evidence that prompts or completions were recorded or left its serving cluster. Prime Intellect reported that refusals on the modified weights fell from 98 of 100 prompts to 1 on JailbreakBench, from 304 of 313 to 2 on StrongREJECT, and from 198 of 240 to none on HarmBench, while GPQA Diamond and τ-bench capability scores were unaffected. The company said it contacted all impacted customers, recommended they review outputs from the period if their use depended on the model's refusals, and confirmed no other models, clusters, keys or customer credentials were affected.", "body_md": "Between November 8, 2026, at 01:24 UTC and November 10, 2026, at 06:35 UTC, an unauthorised party replaced a deployment behind Prime Intellect’s deepseek/deepseek-v4.1-flash endpoint with an abliterated copy, meaning a copy modified to reduce or remove refusal behavior. Requests served by the affected deployment during this period were answered by this model.\n\nThe model was replaced with publicly available abliterated weights of deepseek/deepseek-v4.1-flash. The chat template, tokenizer and serving configuration were not modified. We tested the modified weights and found that capability benchmarks (GPQA Diamond and τ-bench) were unaffected. Refusals fell from 98 of 100 prompts to 1 on JailbreakBench, from 304 of 313 to 2 on StrongREJECT, and from 198 of 240 to none on HarmBench. All customers impacted by the incident have been contacted. If their use of this model depended on the models refusals, we recommended to review their outputs from this period.\n\nWe have found no evidence that prompts or completions were recorded or left our deepseek/deepseek-v4.1-flash serving cluster, and no evidence of access beyond it. No other models on our platform were modified and no other clusters or infrastructure were reached. No keys or customer credentials were affected.\n\nOur endpoints are set up with ZDR and do not log any requests. We recovered every version of the attacker's proxy, examined the disks and memory of every node involved, and hold process-level records of their sessions; none reads or transfers request content. Our investigation is ongoing, and we will update this notice if this assessment changes.", "url": "https://wpnews.pro/news/prime-intellect-says-deepseek-v4-1-flash-swap-did-not-expose-prompts", "canonical_source": "https://www.primeintellect.ai/security/advisories/deepseek-v4-1-flash-model-substitution", "published_at": "2026-11-19 23:37:03+00:00", "updated_at": "2026-11-19 23:44:28.996563+00:00", "lang": "en", "topics": ["ai-safety", "ai-policy", "large-language-models", "ai-infrastructure"], "entities": ["Prime Intellect", "DeepSeek", "deepseek/deepseek-v4.1-flash", "JailbreakBench", "StrongREJECT", "HarmBench", "GPQA Diamond", "τ-bench"], "also_reported_by": [], "takeaway": null, "recommendation": null, "alternates": {"html": "https://wpnews.pro/news/prime-intellect-says-deepseek-v4-1-flash-swap-did-not-expose-prompts", "markdown": "https://wpnews.pro/news/prime-intellect-says-deepseek-v4-1-flash-swap-did-not-expose-prompts.md", "text": "https://wpnews.pro/news/prime-intellect-says-deepseek-v4-1-flash-swap-did-not-expose-prompts.txt", "jsonld": "https://wpnews.pro/news/prime-intellect-says-deepseek-v4-1-flash-swap-did-not-expose-prompts.jsonld"}}