YC F26's Deepmark embeds inaudible IDs in AI agents' voices to verify callers
ycombinator · x · 2026-09-23
Deepmark, launching today out of Y Combinator's F26 batch, embeds an inaudible identifier inside an AI agent's voice so the person on the other end of a phone call can verify whether they are talking to an AI and who is really speaking.
The team also demoed the verification live on a single phone, showing the watermark working in an actual call.
More from Safety
- Anthropic: Fully Automated AI Research Needs Much Stronger Safety Standards — haider1 · 2026-09-23
- AI Security Model dfs-large1 Found TikTok Flaw Exposing Photos and Camera Access — andreamichi · 2026-09-23
- OpenAI tightens RL environment filtering, citing flawed environments as source of misalignment — ShakeelHashim · 2026-09-23
- Google's ESCAL8 security conference heads to Singapore with AI agent focus — moyix · 2026-09-23
- Opus 5.5 Is First Opus with Fable 5.1-Class Safeguards on Cyber, Bio, Frontier LLM Dev — ClaudeDevs · 2026-09-23
- Researcher: as model risks escalate, we still have tools we trust today — sleepinyourhat · 2026-09-23