What's trending in AI right now, from the app charts to the community feeds. Real links, our take.
ChatGPT just got a lot more visual
OpenAI rolled out GPT-6 and a new feature called Intelligent UI this week. Paid users got it on October 7 and free accounts followed the next day. ChatGPT responses can now include interactive elements built right into the conversation: tappable diagrams, live calculators, recipe widgets, editable charts, and maps, rather than text alone. MacRumors has screenshots of what it looks like across several different question types. The practical change is that when you ask about something numerical or visual, you are more likely to get something you can actually use on screen rather than paragraphs to read through and interpret yourself.
Anthropic released a smaller, much cheaper model
Claude Haiku 5.5 launched on October 7, completing Anthropic's current Claude 5.5 lineup. It is designed for high-volume, fast tasks rather than deep reasoning: summaries, classifications, quick lookups, the kind of work that needs to run cheaply at scale. Compared to the previous Haiku model, it costs roughly 90 percent less per token at the lowest pricing tier. SiliconAngle reports that Anthropic also halved Sonnet 5.5 cache read prices on the same day. For regular users, cheaper small model inference tends to show up as more AI features at the same subscription price inside apps that use Claude behind the scenes.
AI video creation apps are climbing across the charts
Several AI video generator apps are moving up at the same time across the App Store's photo, video, and entertainment categories this week. Apps like Vixel, Vidmor, Pose, and VibeShort are all in the mix. Zapier's current list of AI video tools runs to 16 entries, a considerably longer list than a year ago. Going from a text prompt to a short clip entirely on your phone has gotten fast enough and good enough that people who might have skipped these apps a few months ago are trying them now, which is probably why you are seeing more AI-generated video turning up in social feeds.
ChatGPT's memory is frustrating a growing number of users
ChatGPT's memory feature is supposed to save details across conversations so the app remembers things like your job, ongoing projects, or preferences without you needing to repeat them every session. This week, multiple threads across Reddit and the OpenAI community forum are rising around the same complaint: memories the app confirmed it saved are not appearing in new conversations, or the model is silently ignoring them. One thread describes it as a productivity regression for anyone who depends on saved context. OpenAI has not posted a public update on a fix as of this writing.
Can your employer see your ChatGPT prompts?
This question has been showing up frequently in AI communities this week, often from people using a personal ChatGPT account at work. A current guide to the question puts it plainly: on a personal device, personal network, and personal account, your employer almost certainly cannot see what you type. On a work-issued laptop, the situation changes. Monitoring software can capture keystrokes regardless of which account you are signed into. The rule that security advisors tend to land on is the same one that applies to work email: assume anything typed on a work device could be visible.