OpenAI and Anthropic both speak at once with dueling voice updates
The New Stack Amanda Caswell ● Covered by 8 sources
OpenAI and Anthropic both dropped big voice updates the same day, but with totally different goals. One wants your voice to control your computer, the other wants it to help you think.
Based on reporting by The New Stack, Amanda Caswell — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Thursday turned into an unplanned showdown between OpenAI and Anthropic, both pushing significant voice upgrades within hours of each other. Neither company mentioned the other, but the timing wasn't subtle, and the split in strategy says a lot about how each lab sees the future of talking to machines.
OpenAI's move is the bigger swing. ChatGPT Voice, now paired with GPT-Live, is rolling out to the desktop app on macOS and Windows for Plus, Pro, Business, Enterprise, and Education users. The pitch: stop typing, start commanding. Launch it with a keyboard shortcut, keep working normally, and let ChatGPT orchestrate tasks through ChatGPT Work and Codex in the background. On macOS specifically, a new feature called Appshots gives the assistant a look at whatever app you're currently in, so it has context before it acts. Sessions can now juggle multiple tasks at once too, meaning you can fire off a second request while the first is still running instead of babysitting one task at a time. ChatGPT Remote is already live on iOS, with Android reportedly on the way.
Anthropic went the opposite direction. Rather than turning voice into a remote control, Claude's updated Voice Mode is built for slower, more deliberate back-and-forth — the kind of conversation where you're working through a gnarly problem out loud and want the model to actually pause and think before jumping in. Claude Code Voice picked up dictation support too, letting developers speak prompts, terminal commands, and code edits either continuously or through push-to-talk. It's less about replacing your hands and more about giving your brain a sounding board.
What's notable isn't just that both companies shipped voice features on the same day — it's that they clearly disagree on what voice is for. OpenAI is betting that speech becomes the interface for controlling software and agents across your whole machine. Anthropic is betting that speech is better suited to reasoning and iteration, not clicking buttons. Both are still just chat products at their core, but the direction each lab is nudging users in reveals where they think the real value sits.
Either way, keyboards just got a little more optional.
My take — AI-written commentary, not fact-checked reporting
I'll say the obvious thing nobody wants to: this is a coincidence of product calendars, not some grand rivalry, but it's still a useful signal. OpenAI keeps chasing the everything-app dream where ChatGPT quietly runs your whole desktop, which is exactly the kind of ambient control that should make people nervous about permissions creep. Anthropic's version is smaller and, frankly, more honest about what voice is actually good for right now — thinking out loud, not driving your OS. I'd rather have a good thinking partner than a assistant with root access to my apps, but I suspect the market will reward whichever one feels more magical in a demo, not whichever one is more sensible.
Read more about this at: The New Stack
