0:00 / 2:59
Chapters
Sources
DAILY ROUNDUP
ChatGPT Gets Voice Control on Desktop, Claude Voice Mode Upgrades, Tongue-Controlled Mouse!
calendar_today Date:
schedule Duration: 2:59
OpenAI brings voice control to ChatGPT desktop, Claude's voice mode gets a model upgrade, Claude Code adds a security scanner, and a tongue-controlled touchpad launches nationwide.
- 01. OpenAI brought ChatGPT Voice to its desktop app, letting you control your computer and direct multiple agents by talking
- 02. Anthropic upgraded Claude's voice mode to run on its more capable models instead of the smallest one
- 03. Claude Code's new Security plugin scans for vulnerabilities before you commit or across a full codebase
- 04. Augmental's MouthPad, a tongue-controlled touchpad, is now available across the whole US
OpenAI has brought ChatGPT Voice to its desktop app, letting users control their computer and direct multiple agents running in ChatGPT Work or Codex simply by talking. The feature is powered by GPT-Live, the same voice model family capable of listening and speaking simultaneously, allowing it to coordinate work across several threads mid-sentence. Users can start a task in voice mode, then check or steer work happening elsewhere without touching the keyboard. It's rolling out globally today and plugs directly into OpenAI's existing Computer Use feature, meaning the voice layer sits on top of tools that already control the machine rather than operating separately.
Anthropic has made a similar move, upgrading Claude's voice mode to run on its more capable models rather than routing every query through the smallest, fastest one. This means users can now talk through genuinely difficult problems out loud and receive the same quality of answer they'd expect from typing, across more languages than before. The upgraded voice mode also reaches directly into tools already connected mid-conversation, so it's no longer a stripped-down version of Claude but the same assistant, just spoken.
Separately, Anthropic has released the Claude Security plugin for Claude Code, now in public beta. It scans for vulnerabilities before code is committed or runs a full audit across an entire codebase. The tool uses a multi-agent setup that maps how the code actually works across files, builds a threat model, then searches for serious issues such as injection flaws, authentication bypasses, memory corruption and logic errors that pattern-matching scanners often miss. Every finding is independently reviewed before appearing in the terminal, and developers still approve every fix themselves, turning security review into something that runs continuously alongside development rather than a task teams get to eventually.
Elsewhere, Augmental's MouthPad, a touchpad controlled entirely with the tongue, is now available across the whole of the US. It fits over the upper teeth like a slim retainer, combining a pressure-sensitive pad on the palate with head tracking to move a cursor, click and swipe, all without the hands leaving whatever the user is doing. Over a hundred people already use one daily, with some wearing it for up to sixteen hours at a stretch. It's designed primarily for people who can't use a traditional mouse, though its potential use cases may extend well beyond that original purpose.
Meta Data
Model:
