LESSWRONG
LW

Dipika Khullar

Message

11mo

We Built a Tool to Protect Your Dataset From Simple Scrapers

Author: Alex Turner. Contributors: Dipika Khullar, Ed Turner, and Roy Rinberg. Dataset contamination is bad for several reasons. Most obviously, when benchmarks are included in AI training data, those benchmarks no longer measure generalization -- the AI may have been directly taught the answers. Even more concerningly, if your data...

Jul 25, 2025•60

Dipika Khullar

11mo

Dipika Khullar — LessWrong

Dipika Khullar

Message

11mo

We Built a Tool to Protect Your Dataset From Simple Scrapers

Jul 25, 2025•60

Dipika Khullar

11mo

We Built a Tool to Protect Your Dataset From Simple Scrapers

TurnTrout

TurnTrout, Edward Turner, Dipika Khullar, Roy Rinberg+ 0 more

TurnTrout, Edward Turner, Dipika Khullar, Roy Rinberg

7mo

Author: Alex Turner. Contributors: Dipika Khullar, Ed Turner, and Roy Rinberg.

Dataset contamination is bad for several reasons. Most obviously, when benchmarks are included in AI training data, those benchmarks no longer measure generalization -- the AI may have been directly taught the answers. Even more concerningly, if your data promote negative "stereotypes" about AIs, they might become self-fulfilling prophecies, training future models to exhibit those very behaviors.

In the Claude 4 system card, Anthropic revealed that approximately 250,000 transcripts from their alignment faking paper had been scraped from the public web and included in their pretraining data. This caused an early model to hallucinate details from the paper's fictional scenarios, forcing Anthropic to... (read 722 more words →)