OpenAI Launches AI Keypad Hardware for Secure Inputs

In a surprise move into the hardware sector, OpenAI has announced the developer 'AI Keypad,' a physical USB-C peripheral designed to enhance local system security.

The keypad is engineered to cryptographically sign terminal inputs, ensuring that autonomous AI agents cannot execute unsanctioned code commands. System admins and developers looking to keep their local workspaces clean can utilize [Code Formatter](/tools/code-formatter/).

Cryptographic Security at the Hardware Interface Layer

By routing model permissions through a physical, hardware-level key, OpenAI aims to neutralize prompt injection threats where models are manipulated into executing malicious shell scripts.

Neutralizing Prompt Injection on Local Developers

The device is currently shipping to select enterprise beta partners, with a wider retail release planned for early 2027.