🍎 Apple tests ChatGPT rival Veritas 🍎
Apple employees are stress-testing Siri's AI overhaul using an internal ChatGPT-like chatbot.

Apple builds ChatGPT-style Veritas app codenamed 'Veritas' that employees use to test advanced features including searching personal data, editing photos with voice commands, and performing in-app actions. The app maintains conversational context and represents Apple's strategic push to compete in the AI assistant space. However, engineering challenges have pushed the AI-powered Siri launch to March 2026, causing executive reshuffles and talent departures. This internal testing tool demonstrates Apple's commitment to overhauling Siri with sophisticated AI capabilities that could rival existing chatbots.

🎬 AI actress attracts Hollywood talent agents 🎬
Synthetic performer Tilly Norwood is actively negotiating with multiple Hollywood agencies.

AI talent studio Xicoia revealed that its AI actress attracts Hollywood talent agents with Tilly Norwood in active negotiations with multiple agencies, marking a potential breakthrough for synthetic performers. The development has sparked significant backlash from actors calling for boycotts of any agencies that sign AI performers. Industry insiders report studios went from dismissing AI talent to actively pursuing deals within just months. This represents a crucial test case for how far the entertainment industry's AI transformation will extend, potentially disrupting traditional talent representation models and raising questions about the future of human performers.

📈 Anthropic researcher dismisses AI slowdown claims 📈
Evidence suggests AI capabilities continue doubling at exponential pace despite recent speculation.

Anthropic's Julian Schrittwieser presents compelling evidence that Anthropic researcher dismisses AI slowdown claims as AI capabilities continue doubling at a steady exponential pace. A new METR study demonstrates that AI models can now handle software engineering work autonomously for over two hours with a 50% success rate—double the performance achieved just seven months ago. If current trends continue, the research suggests models could operate independently for full eight-hour workdays by mid-2026, indicating that AI development remains on track for significant capability improvements rather than hitting scaling limits.

INTERESTING TO KNOW

🎬 OpenAI builds TikTok-style app for AI videos 🎬

OpenAI developing TikTok-style video app powered by its upcoming Sora 2 video model, designed to mimic TikTok's vertical scrolling interface but featuring exclusively AI-generated content. The platform will limit video clips to 10 seconds and include identity verification systems. This represents OpenAI's first major foray into social media, potentially creating a new category of AI-native social platforms while raising questions about content authenticity.

🕵️ North Korea uses AI tricks 🕵️

CNN investigation uncovered thousands of North Korean workers attempting to secure remote US employment under stolen identities, using sophisticated AI tools including ChatGPT, face-swapping software, and live face-masking technology to hide their identities during video interviews. These operatives successfully infiltrated companies like Nike and Bank of America while funneling millions in salaries back to the North Korean regime.