AI News

Cancerous Backdoor Attack: invisible tokens plus a stage controller to corrupt LLM reasoning over time

ABC2 min read
Research