ChatGPT’s upgraded voice mode is better at shutting up
THE SO WHAT
Turn‑taking and interruption control are becoming competitive features—voice agents are being judged by conversational etiquette, not just IQ. For any voice interface you own, UX details like barge‑in, pauses, and “knowing when to be quiet” now matter as much as model choice.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIGoogle Photos adds a new AI ‘Video Remix’ tool
Consumer-grade relighting and background swaps in Google Photos push pro-level video editing into the default camera roll. If you’re building creative tools, assume your baseline user now expects cinematic effects as a one-tap filter, not a pro workflow.
Applied AIExperts warn of the 'first documented case of agentic ransomware' — dangerous JADEPUFFER attack run entirely by an LLM
Agentic ransomware run entirely by an LLM—badly enough that it loses its own encrypted data—shows offensive automation is arriving before reliability. Treat LLM access, tool APIs, and prompt channels as part of your attack surface and start logging and rate-limiting them like you would any untrusted code execution.
Applied AIGrok 4.5 is available in Grok Build, in Cursor, and from the SpaceXAI console, but not in the EU; it costs $2 per 1M input tokens and $6 per 1M output tokens
Grok 4.5 coming in at $2 / 1M input and $6 / 1M output tokens with tight integration into Cursor is a direct price–performance shot at enterprise coding, legal, and finance workloads. If you’re heavy on structured text work, it’s time to benchmark Grok 4.5 against your current model and re-run your unit economics—especially outside the EU.
Applied AIWhy this CEO thinks video games make better training data than the internet
Treating video games as structured, physics-rich training data is a bet that synthetic, controllable worlds will beat scraped internet text for AGI-relevant learning. If you operate in gaming or simulation, there’s a new monetization lane: your environments as data and curriculum, not just entertainment.