
Image via TechCrunch
Thursday, January 22, 2026
Apple's turning Siri into ChatGPT
Adobe's turning PDFs into podcasts (yikes, the reading is officially over), while Apple's making a bold move to transform Siri into a ChatGPT-like chatbot for iOS 27. Meanwhile, Google's weaving Gemini directly into Chrome with multi-tab context and reasoning chops, Anthropic published their Claude constitution for transparent AI values, and here's the wild part: they're redesigning hiring tests faster than Claude can beat them. So real talk: if your AI can optimize its way past every test we design, what exactly are we measuring anymore?

Image via TechCrunch
Top Stories
Adobe adds podcast and presentation generation to Acrobat, letting users transform PDFs into audio summaries and visual decks through AI, addressing the reality that many people skip reading lengthy documents.
TechCrunch
Apple is converting Siri into a ChatGPT-style chatbot with voice and text capabilities, signaling a strategic pivot driven by competitive AI pressures and partnerships with Google's Gemini to catch up in the generative AI race.
Anthropic
Anthropic published Claude's new constitution, a comprehensive values document designed to shape model behavior through principle-based reasoning rather than rigid rules, released publicly to promote transparency and enable community critique of AI alignment approaches.
Advanced reasoning models achieve superior performance through internal multi-agent-like interactions and perspective diversity rather than extended computation alone, suggesting that social organization of thought is key to sophisticated AI reasoning.
As AI models rapidly improve, traditional technical assessments become obsolete; Anthropic reveals how they redesigned hiring tests multiple times as successive Claude versions defeated them, ultimately pivoting to unusual constraint-based puzzles where human reasoning still prevails.
Keep Reading
Industry Voices
Nathan Lambert
Research Scientist at Allen Institute for AI (AI2)
Deep dives into RLHF mechanics and post-training methods that actually shape how chatbots behave, with refreshingly candid takes on what works and what's hype.
Carlos Guestrin
Researcher at Stanford
Pioneer of interpretable ML and tree-based models who co-created XGBoost and founded the team behind Apple's Core ML recommendations.
Enjoyed this issue?
Get daily AI intel delivered to your inbox. No fluff, just the stories that matter.