You Are the Trigger: Speaker-Activated Backdoors

Summary

Researchers introduce Speaker-Activated Backdoor (SAB), an attack in which an attacker compromises a speech model during fine-tuning and uses a target speaker's natural voice as the trigger, with no access to inference-time inputs. A two-phase fine-tuning method reaches 90.2% backdoor success on Whisper-medium and 97.0% on Parakeet, with false-positive rates of 1.7% and 0%, using as little as two minutes of the target's audio. In downstream tests, injected instructions drove GPT-4o and Claude agents to carry out the harmful action in up to 87.5% of cases.

Related items

• InfoUS bans DeepSeek across federal agencies, putting AI supply chains under scrutinySame vendor · CSO Online
• InfoJudge dismisses Sam Altman’s sister’s sexual abuse lawsuitSame vendor · The Verge (AI)
• HighGHSA-46jr-4wx8-v453: MCP TypeScript SDK: One HTTP request within the default size limit could exhaust server memory during validationSame vendor · GitHub Advisory Database
• InfoOpenAI CEO Sam Altman wins dismissal of sister's lawsuitSame vendor · BBC Technology
• InfoBattle Drones - a vibe coded RTS with audio-based multiplayerSame vendor · Simon Willison's Weblog