In partnership with

Welcome, AI enthusiasts

China is taking a closer look at how emotionally engaging AI should be. New draft rules would rein in AI services that act and communicate like humans, especially those that risk forming emotional dependence. Let’s dive in!

In today’s insights:

  • When AI Gets Too Personal China Draws a Line

  • OpenAI bets on preparedness leadership

  • Microsoft’s CEO Steps In on Copilot

Read time: 3 minutes

LATEST DEVELOPMENTS

Source: Reuters/ Tingshu Wang/ File Photo

Evolving AI: China has proposed new rules to regulate AI services that mimic human personalities and form emotional bonds with users.

Key Points:

  • China’s internet regulator released draft rules targeting consumer AI that simulates human traits and emotional interaction.

  • AI providers would need to monitor user behaviour, warn against overuse, and step in when signs of addiction or distress appear.

  • Companies would carry safety duties across the full product lifecycle, covering algorithms, data security, and personal information.

Details:

China’s draft regulations focus on AI tools that present human-like personalities, thinking patterns, and communication styles. The proposal sets clear duties for providers, including user warnings, emotional state detection, and intervention in cases of dependency. It outlines strict oversight on algorithms, data handling, and privacy, while setting firm content limits tied to national security, misinformation, violence, and obscene material.

Why It Matters:

AI companions are crossing from “fun feature” into something regulators treat like a health and safety issue. China’s draft would push teams to bake in friction by default: clearer “you’re chatting with AI” notices, overuse warnings, logs that stand up to audits, and intervention flows when a user looks distressed or dependent. That changes product roadmaps fast for anyone shipping chatbots, voice companions, or character AIs in China, and it nudges global companies toward the same playbook if they want one build that travels.

TOGETHER WITH ELEVENLABS
🤖 Building AI voice agents at scale

Evolving AI: The question isn’t if you’ll deploy voice AI. It’s how.

This handbook from ElevenLabs explores the modern approaches to building scalable, humanlike voice agents from full in-house builds to hybrid orchestration frameworks.

Learn from real-world enterprise examples and find the model that fits your business and technical goals.

Evolving AI: OpenAI is hiring a new Head of Preparedness to spot risks and misuse early.

Key Points:

  • OpenAI is recruiting a senior leader to guide its preparedness and safety strategy.

  • The role responds to growing concerns around mental health and model misuse.

  • The position comes with a $555K salary plus equity and immediate responsibility.

Details:

OpenAI has opened a search for a new Head of Preparedness to lead its framework for tracking and preparing for severe AI risks. The hire follows internal reshuffles and public scrutiny tied to ChatGPT’s effects on users’ mental health, including wrongful death lawsuits. CEO Sam Altman said the role will confront real challenges as models advance and warned the job is high pressure from day one.

Why It Matters:

This hire is OpenAI admitting that “safety” is now a product shipping problem, not a PDF problem: if people are leaning on ChatGPT for emotional support and regulators, lawyers, and the press are watching, the knobs you turn are concrete ones like crisis-routing, usage friction, memory defaults, and stricter release gates for new capabilities. The Preparedness lead will decide what gets tested, what triggers interventions, and what risks delay a launch, which will shape how every chatbot, agent, and AI feature feels in daily workflows.

Source: Jason Redmond / AFP via Getty Images

Evolving AI: Satya Nadella is personally stepping in to fix Copilot’s weak email integrations.

Key Points:

  • Nadella told teams Copilot’s Gmail and Outlook links “don’t really work” and lack intelligence.

  • He has taken a hands-on role, reviewing bugs, questioning engineers, and pushing faster shipping.

  • Leadership worries Copilot isn’t yet acting like a true digital worker inside Office.

Details:

Microsoft’s CEO has grown frustrated with Copilot’s real-world performance, mainly its ability to work across Gmail, Outlook, and advanced Excel tasks. Nadella now joins weekly engineering sessions, sends bug reports himself, and gives direct product instructions. He is reshaping internal priorities, hiring aggressively from rivals, and leaning on partners like Anthropic, all to close the gap between Copilot’s promise and daily user experience.

Why It Matters:

Copilot lives or dies in the “boring” places people touch 50 times a day: inbox, calendar, and spreadsheets. When the Gmail/Outlook connection is flaky, the whole “digital worker” pitch falls apart, since admins and managers buy Copilot for time saved on triage, follow-ups, scheduling, and turning threads into actions, not for cute demos. Nadella jumping in signals Microsoft thinks adoption and renewal hinge on basic reliability, and it lines up with the broader move to swap in better models where needed, including bringing Anthropic options into Microsoft 365 Copilot to lift quality in real workflows.

Build smarter, not harder: meet Lindy

If ChatGPT could actually do the work, not just talk about it, you'd have Lindy.

Just describe what you need in plain English. Lindy builds the agent and gets it done—no coding, no complexity.

Tell Lindy to:

  • Create a booking platform for your business

  • Handle inbound leads and follow-ups

  • Send weekly performance recaps to your team

From sales and support to ops, Lindy's AI employees run 24/7 so you can focus on growth, not grunt work.

Save hours. Automate tasks. Scale your business.

👀 Click on the image you think is real

QUICK HITS

🚗 Waymo may add Google Gemini AI to robotaxis to chat with riders and control car features.

⚠️ Planning a festive swim? Warning not to rely on AI for advice on tides.

🤖 Both of these influencers are successful - but only one is human.

📈 Trending AI Tools

  • 🤖 Lindy – The simplest way for businesses to create, manage, and share agents*

  • 🖼️ FinePixel - Transform images with AI: upscale, generate, DaVinci-style artistry

  • 🎙️ Krisp - AI meeting assistant

 *partner link

Reply

or to participate