01 Google Photos builds a virtual wardrobe for AI‑powered try‑ons
Google is launching an AI-powered try-on feature in Google Photos that creates a virtual “wardrobe” from clothing in a user’s gallery, allowing people to mix and match outfits, save looks, and share them, according to reporting in The Verge.
The Verge describes a Google-shared demo video that shows the tool using existing photos to render clothes on different body poses and combine items into new outfit configurations; Google positions the feature as a way to explore looks from clothing you already own.
The feature surfaces productized flows—saving looks and sharing with friends—rather than presenting the capability only as a research demo, signaling Google’s intent to integrate AI image-editing into consumer workflows.
Separately, TechCrunch’s coverage of Runway and the broader AI-video sector underscores that image and video generation tools are evolving quickly; Runway’s CEO framed video models as a step toward larger “world models,” illustrating how advances in generative media are informing product roadmaps across the industry.
- Google Photos will create a virtual wardrobe from photos in your gallery and let you mix, save, and share outfits (The Verge).
- Google demonstrated the tool in a video that shows rendering clothes on different poses and composing outfits.
- The launch signals Google’s push to embed generative-image features into everyday consumer apps, while the wider industry moves toward larger, multimodal models (TechCrunch).
02 OpenAI proposes a five‑part plan to harden cybersecurity with AI
OpenAI published a blog post outlining a five-part action plan aimed at strengthening cybersecurity in what it calls the Intelligence Age, emphasizing broader access to AI-powered defensive tools and protective measures for critical systems.
The plan lays out measures that include democratizing AI tools for cyber defense, improving detection of misuse, and coordinating with industry and public stakeholders to protect infrastructure, according to OpenAI’s post.
OpenAI frames the effort as proactive work to balance the benefits of AI for defenders while limiting actors who would exploit model capabilities for cyberattacks.
The company’s recommendations reflect a push for collaborative, system-level responses to risks that arise as AI capabilities become more powerful and more widely available.
- OpenAI published a five-part cybersecurity action plan focused on democratizing AI defense and protecting critical systems (OpenAI Blog).
- The plan emphasizes misuse detection, coordination with external stakeholders, and tools to help defenders leverage AI.
- OpenAI presents the measures as part of a broader effort to manage security risks as AI capabilities scale.
03 Google marks 20 years of Translate with new tips and feature highlights
Google published a post celebrating 20 years of Google Translate that highlights product tips, new features to try, and illustrated examples of how users and developers employ the service, according to the Google AI Blog.
The post showcases use cases such as live conversation translation and other consumer-facing features, and it frames the anniversary as a moment to surface practical guidance for existing Translate users.
By emphasizing tips and feature highlights rather than a single major technical announcement, Google appears to be positioning Translate’s milestone as both a user-facing anniversary and a chance to remind people of the product’s current capabilities.
The blog entry ties Translate’s consumer features to ongoing investments in language technology across Google’s products.
- Google’s AI Blog published a 20th-anniversary post for Translate, including tips and feature highlights (Google AI Blog).
- The post emphasizes live conversation translation and practical guidance for users.
- Google frames the milestone as both a consumer-facing anniversary and a prompt to showcase recent language-tech capabilities.
What moved around the edges
OpenAI restates ChatGPT community‑safety commitments
OpenAI published a post describing how it protects community safety in ChatGPT through model safeguards, misuse detection, policy enforcement, and collaboration with safety experts.
OpenAI BlogChatGPT downloads slow as uninstall rates climb
The Verge reported that ChatGPT’s downloads are slowing and uninstalls have increased year over year, a trend that analysts say could complicate OpenAI’s growth metrics ahead of a potential IPO.
The Verge AIParallel Web Systems hits a $2B valuation after new funding
TechCrunch reported that Parallel Web Systems, an AI agent-tool startup founded by Parag Agrawal, raised $100 million led by Sequoia and reached a $2 billion valuation just months after an earlier raise.
TechCrunch AIDHS moves to expand purchases of Predator‑style MQ‑9 drones
Reporting shows U.S. Customs and Border Protection is spending hundreds of millions of dollars on additional high-end surveillance MQ-9-style drones, and other DHS components may expand their own fleets.
404 Media AIGitHub to bill Copilot customers based on actual AI usage
GitHub announced it will begin charging Copilot users based on measured AI usage to offset rising inference costs, signaling a shift from flat pricing toward usage-based billing.
Ars Technica AI