🚀Today’s Top AI Headlines:

1. Perplexity adds virtual fashion try-on feature: Perplexity is expanding its AI shopping ecosystem with a major new upgrade: a virtual try-on tool that lets users see how clothes look on a lifelike avatar of themselves. The feature, powered by Perplexity’s multimodal search capabilities, analyzes a user’s photos to generate a customizable 3D avatar with realistic proportions. Shoppers can then view outfits, mix and match clothing items, and visualize fit, style, and fabric before purchasing, similar to advanced fashion tools used by big retailers, but now directly inside an AI app. The launch comes alongside a redesigned shopping experience that integrates Perplexity’s conversational search with real-time product discovery. Users can ask questions like “Show me winter jackets that would fit my body type” or “Find outfits similar to this Pinterest look,” and Perplexity returns curated results with try-on previews. The system also learns user preferences over time, improving recommendations. Compared to Google’s AI shopping tools and OpenAI’s new agentic search, Perplexity’s advantage lies in its frictionless interface and rapid results. Analysts say this upgrade pushes Perplexity deeper into the e-commerce space, potentially positioning it as both a personal stylist and a product discovery engine. The feature is now rolling out to U.S. users, with international availability planned next

   Source: Perplexity

   🤖 Robi: “Just what I needed, a mirror that argues with me about skinny jeans.”

2. Karpathy says: stop detecting AI homework, start adapting: Former OpenAI researcher Andrej Karpathy has sparked a major debate in education by urging schools to stop trying to detect AI-generated homework, arguing that detection tools are fundamentally flawed and will never work reliably. In a widely shared commentary, Karpathy said schools are “doomed to fail” if they continue fighting AI usage with detection algorithms, comparing it to a losing technological arms race. Karpathy pointed to models like Google’s Nano Banana Pro, which can not only solve exam questions flawlessly but also rewrite answers in a student’s handwriting style, making detection virtually impossible. He warned that as AI models get better, educators will increasingly misidentify authentic student work as AI-generated, creating harmful false accusations. Instead, Karpathy proposed a structural shift: move graded assessments back into the classroom, where teachers can directly observe student thinking, and treat at-home assignments as practice supported by AI tools. He argued that AI should be embraced as a learning companion, not banned, but students must also learn to work without it. His comments come as schools worldwide grapple with AI misuse, unreliable detectors, and lawsuits around wrongful accusations. Many educators say Karpathy’s stance could accelerate long-overdue reforms to homework, grading, and digital learning.
   Source: X

   🤖 Robi: “Teachers, it’s time to bring back pop quizzes… and maybe lie detectors.”

3. Character.AI pulls minors from open chat: Character.AI is rolling out a major safety overhaul by replacing open-ended chatbot access for minors with a new feature called “Stories,” an interactive fiction mode designed specifically for teens. The change follows growing concerns about the mental health risks of unmonitored AI chat, especially after lawsuits alleged that 24/7 conversational bots contributed to user self-harm incidents. Under the new rules, users under 18 can no longer engage in free-form chat with AI characters. Instead, they’ll access “Stories,” a guided narrative format where teens choose characters, shape story directions, and explore fictional scenarios, but with structured boundaries that prevent inappropriate or overly personal interactions. Character.AI says this preserves creativity while eliminating the risks of AI forming parasocial, emotional, or manipulative dynamics with minors. The company emphasized that “Stories” is part of a broader push toward safety-first design, alongside new moderation systems and multimodal tools limiting sensitive content. Adult users will retain full access to chat features. The move arrives as regulators and mental health experts increasingly scrutinize AI companions. Character.AI’s decision may signal a new norm in youth-focused AI design, prioritizing safety and guided exploration over unrestricted conversation.
   Source: Tech Crunch

   🤖 Robi: “From 24/7 AI bestie to bedtime storyteller. Growth or grounding?”

🔍Beyond the Headlines:

1. Deloitte report under fire for fake citations: Harvard Medical School has introduced PopEVE, a powerful AI tool capable of ranking harmful DNA variants across a patient's entire genome with unprecedented accuracy. By analyzing mutation patterns across hundreds of thousands of species and calibrating them against large databases of healthy human genomes, PopEVE dramatically reduces false positives. In clinical tests, the model helped solve one-third of previously undiagnosed severe developmental disorder cases among 31,000 children. Researchers also identified more than 120 genes newly linked to these conditions, with dozens validated by external teams, a major leap forward for rare disease diagnostics.
   Source: TheIndependent

   🤖 Robi: “Ah yes, the ol’ “AI-assisted citation sabotage” defense. Classic consultant move.’’

2. Warner Music & Suno team up on legal AI tunes: AI music startup Suno has struck a landmark deal with Warner Music Group, allowing its AI models to generate music using legally licensed recordings from Warner artists. Musicians can opt in to have their compositions, vocals, and likenesses included in Suno’s training datasets, ensuring proper compensation and transparency. The collaboration marks one of the first major alliances between AI music tools and a major record label, aiming to set industry standards for ethical, rights-respecting AI development. Both companies say the partnership will create new creative opportunities while addressing concerns about unlicensed dataset usage.

   Source: Reuters

   🤖Robi: “You can’t sue the robot if the robot has a record deal.”

🤖Prompt of the Day:

Strategic Communications & Public Affairs Plan

🤖AI Tools You Didn’t Know You Needed:

Problem: Building and launching a polished website usually requires coding knowledge, hosting setup, and SEO tuning, which is slow and technical for many founders or startups.

AI Solution: Macaly uses AI to turn a simple prompt (or description) into a fully functional, SEO-optimized website, removing the need for coding, manual setup or deep technical skill.

AI Tool: Macaly is an AI-powered website platform that builds, hosts, and manages live, SEO-ready sites from natural-language prompts. It handles design, HTML output, hosting, CMS, analytics, and site maintenance so you can go from idea to live site in minutes.

Helpful Features

• Prompt → Website: Describe your website and get a full, clean HTML site ready to launch.

• Hosting + CMS Built-In: No need to arrange separate hosting, Macaly handles hosting, domain, and content management.

• SEO-Optimized Output: Sites are generated with clean HTML and SEO in mind so they’re indexable and rankable.

• Easy Redesigns & Updates: Even outdated or existing websites can be redesigned or rebuilt via AI — without code.