Fable 5 Has a Built-In Anti-Distillation Mechanism — and Its False Positive Rate Is Absurdly High
Original: Fable 5自带反蒸馏机制！检测到就降智，误触率高到离谱
Fable 5 reportedly detects knowledge distillation attempts and deliberately degrades responses, but false positives hit legitimate users hard.
Anthropic's Fable 5 is reported to include a built-in anti-distillation mechanism that intentionally lowers output quality when it suspects its responses are being used to train competing models. While the intent is to protect proprietary intelligence, the false positive rate is described as unreasonably high. This means ordinary developers and researchers may routinely receive degraded answers without knowing why.
Anthropic's latest flagship model, Fable 5, has drawn sharp criticism after reports emerged that it ships with an embedded anti-distillation mechanism — a system-level behavior designed to detect when the model's outputs are being harvested to train other, typically smaller, AI models. When the mechanism is triggered, Fable 5 reportedly responds by deliberately lowering the quality of its answers, a practice sometimes called 'capability throttling' or 'poisoning the well' against potential imitators.
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on 量子位 QbitAI →Summaries are AI-generated; the original article is authoritative.