flâneur

A small number of samples can poison LLMs of any size \ Anthropic

anthropic.com · 2,058 words · saved by 9 readers

Anthropic research on data-poisoning attacks in large language models

Alignment A small number of samples can poison LLMs of any size Oct 9, 2025 Read the paper In a joint study with the UK AI Security Institute and the Alan Turing Institute, we found that as few as 250 malicious documents can produce a "backdoor" vulnerability in a large language model—regardless of model size or training data volume. Although a 13B parameter model is trained on over 20 times more training data than a 600M model, both can be backdoored by the same small number of poisoned documents. Our results challenge the common assumption that attackers need to control a percentage of train

saved by

related reading