New features and launches in ChatGPT, Cursor, Perplexity and other AI tools, plus changes to pricing and availability.
Microsoft just added AI-generated themes to Outlook for Copilot Pro and business subscribers. Users can build custom backdrops inspired by over 100 locations or dynamically updated based on local weather and location, rendered in realistic, oil painting, or cartoon styles. It gives everyday inboxes a neat, personalized touch.
Krea AI just rolled out personalized LoRA training, letting anyone train custom characters, art styles, or products with as few as three uploaded images. The cloud-based setup skips the usual local GPU hassle, making fine-tuning beginner-friendly. Access is rolling out via 100 early invite codes, while regular use requires a paid membership.
Google and Sourcegraph just shared test results from plugging Gemini 1.5 Flash's 1M-token context into the Cody coding assistant. The massive context dropped hallucination rates from 18.97% down to 10.48% on complex codebase tasks, while custom caching trimmed initial response latency from 40 seconds to roughly 5.
Apple just released the iOS 18.2 public beta, opening up the second wave of Apple Intelligence features to public testers. Users can now query ChatGPT directly through Siri without needing an account, and iPhone 16 owners get Visual Intelligence camera search. Image Playground and custom Genmoji are also included, though creative tools still sit behind a secondary waitlist.
Microsoft just rolled out new AI tools to Paint and Notepad for Windows Insiders. Paint gains generative fill to insert new objects via text prompts alongside generative erase, while Notepad adds a Rewrite feature to change tone and length. The catch: generative fill is currently limited to Snapdragon-powered Copilot+ PCs.
Hume AI just launched a browser-based voice application combining its proprietary EVI 2 speech model with Anthropic's Claude 3.5 Haiku. Users can chat with various specialized AI personas through their microphone without creating an account. Early testing shows solid factual answers, though narrative modes still suffer from occasional repetition and voice shifts.
Microsoft just rolled out its new "Xbox Support Virtual Agent" to Xbox Insiders in the US, giving gamers an animated AI assistant to handle console and game support queries. It marks Xbox's first consumer-facing AI test ahead of a broader push across the platform.
Amazon just launched X-Ray Recaps on Prime Video, a generative AI tool that creates text summaries of full seasons, episodes, or even specific scenes. It analyzes video and dialogue to highlight key events while enforcing guardrails to prevent spoilers. The beta is live on Fire TV for its original series.
Apple just introduced ChatGPT usage tracking and Plus upgrade shortcuts in iOS 18.2 Beta 2. Free users get a small daily allowance of GPT-4o requests and two DALL-E 3 images before Siri downgrades to a basic model. If you need more headroom, the Settings app now lets you sign in or subscribe directly to OpenAI's $19.99 monthly Plus tier.
NVIDIA just launched a new AI Blueprint for video search and summarization, enabling developers to build visual AI agents that analyze massive video archives and live streams using natural language prompts. Powered by vision-language models, the workflow cuts months off deployment for smart cities, warehouses, and traffic management.
BFL just released FLUX.1 Tools, adding inpainting, outpainting, depth guidance, canny edges, and variation generation to its text-to-image models. The models are available as open-access weights and API endpoints, though Depth and Canny were later deprecated. It gives creators far more precision over generated scenes.
Perplexity just launched an Election Information Hub to help voters navigate key issues, polling logistics, and live race results. The platform taps into data from The Associated Press and Democracy Works' API, offering source-backed summaries on candidates and ballot measures. It is a bold test of whether AI search can handle high-stakes civic info without hallucinating.
Amazon just delayed its generative AI-powered Alexa revamp into 2025, quietly shuttering its beta program in the process. Testers reported the new assistant was slow, hallucinated, and struggled with basic smart home commands like setting kitchen timers or turning off lights. Why it matters: Building a conversational bot is one thing, but reliably connecting an LLM to legacy smart home hardware is proving to be a massive headache.
Perplexity is quietly testing an in-platform shopping feature called "Pro Shop" that lets users research and buy products directly without leaving the site. Eligible items feature a "Buy with Pro" button with promised free shipping, turning the AI search engine into a full checkout loop.
ElevenLabs just rolled out an open-source project called X to Voice alongside a new Voice Design API. The tool scans any X profile's bio and posts to craft a prompt, generating three unique voice previews and even animating profile pictures with matching lip movements. Developers can now also build custom voices for their AI characters via the API.
Anthropic just launched official Mac and Windows desktop apps for Claude in public beta, alongside a new voice dictation feature on iOS, Android, and iPadOS. The dictation tool lets users record up to 10 minutes of audio that Claude then transcribes and answers. It isn't a live, conversational voice mode like ChatGPT's, but it makes firing off complex prompts on the go a whole lot easier.
Anthropic just launched official Mac and Windows desktop apps for Claude in public beta, available to both free and paid users. The dedicated app lets users invoke the assistant via keyboard shortcuts for focused work. Alongside the desktop rollout, mobile versions also gained a native voice dictation feature supporting recordings up to 10 minutes.
Google just rolled out generative AI capabilities for Google Maps Platform and Google Earth, allowing developers to ground LLMs in real-world data from 250 million places. Google Earth is also getting Gemini-powered multi-step reasoning to slash urban planning analysis time from days to minutes. Next month, Rivian will begin rolling out AI place summaries to its vehicle screens.
OpenAI just built web search directly into ChatGPT, serving up real-time answers with clickable source citations. Instead of sending you off to a traditional search engine, it blends conversational chat with live sports scores, weather, and breaking news. Why it matters: It turns ChatGPT into a full-blown search engine alternative backed by official publisher partnerships.
OpenAI just rolled out its Advanced Voice Mode to ChatGPT desktop apps on macOS and Windows, bringing human-like, low-latency audio conversations directly to your computer. Users can chat back and forth while coding or brainstorming over uploaded files with five new voices. It is currently limited to paid Plus and Team subscribers, excluding the EU and UK.
