In partnership with
Welcome, AI enthusiasts
Seedance 2.0 just pushed AI video into film territory. ByteDance’s new model turns simple inputs into multi-shot cinematic scenes with sound, and early access clips out of China are already forcing serious conversations in film and ad production. Let’s dive in!
In today’s insights:
Hollywood's AI problem is officially here
ChatGPT Ads Arrive
Asking AI About Your Symptoms Might Be Risky
Read time: 5 minutes
LATEST DEVELOPMENTS
Evolving AI: ByteDance debuts Seedance 2.0 bringing long-form cinematic AI video to China.
Key Points:
Seedance 2.0 creates up to two-minute 1080p videos from a single prompt with synced audio.
Multi-shot storytelling keeps characters, motion, and sound consistent across scenes.
Early users say quality now rivals top global AI video systems in realism and flow.
Details:
AI video just hit a new level. ByteDance, TikTok’s parent company from China, has rolled out Seedance 2.0, an AI model that can generate multi-scene cinematic video with native sound, smooth motion, and consistent visuals from basic prompts. Early demos are up online and industry watchers are saying this could disrupt traditional film production workflows. It supports native multi-shot generation, phoneme-level lip sync across multiple languages, and joint audio-video creation, meaning dialogue, sound effects, and music are produced together.
Why It Matters:
Hollywood has been treating AI video like a VFX add-on. Seedance 2.0 flips it into a full production engine. If a prompt can output a rough cut with shots, continuity, lip-synced dialogue, music, and SFX, then the expensive middle layer gets squeezed first. Previs teams, stock shoots, B-roll days, temp edits, even parts of post become optional for a lot of work. The winners won’t just be studios with budgets. It’ll be whoever can move fastest, iterate best, and “direct” a model like a crew. And once viewers can’t tell what’s real, the business side gets messy fast. Rights, consent, training data, and what “authentic” even means will turn into a daily fight.
TOGETHER WITH DATADOG
🔐 The Future of AI, LLMs, and Observability on Google Cloud
Evolving AI: Google’s 7 predictions on AI, LLM, and Observability.
Read the 7 key takeaways from Google’s Director of AI and Datadog’s VP of Engineering as they break down their predictions of the future:
Smarter AI and LLM strategies for your org
Building customer trust in AI outputs
Scaling your tooling as LLM expertise grows
OPENAI
🧾 ChatGPT Ads Arrive
Evolving AI: Ads are now appearing in ChatGPT for free and Go users in the U.S.
Key Points:
Ads in ChatGPT are now showing for U.S. users on Free and Go plans.
Sponsored content appears below responses and is matched to query topics.
Plus, Pro, Business, Enterprise, and Education plans remain ad free.
Details:
OpenAI has started showing ads inside ChatGPT responses for U.S. users on the Free and $8 Go tiers. The rollout was quiet, with sponsored placements clearly labeled and separated from answers. Ads are matched to conversation topics like food or travel. OpenAI says advertisers never see chats, only aggregated metrics. Users can hide ads, manage targeting, or upgrade to remove them.
Why It Matters:
People aren’t just chatting with AI anymore, they’re starting to browse inside it. Ads in ChatGPT turn everyday questions into moments of commercial intent, closer to Google Search than social feeds. For users, the choice becomes clear: pay with money or pay with attention. For brands, this creates a new surface where curiosity, decision making, and purchases sit in the same place. The real tension is whether people keep trusting answers once promotion sits nearby, or if they quietly change how and where they ask questions next.
AI IN RESEARCH
⚠️ Asking AI About Your Symptoms Might Be Risky
Source: Getty Images
Evolving AI: Study warns AI medical advice can mislead patients and miss urgent care.
Key Points:
Oxford-led research found AI chatbots give mixed medical guidance users struggle to judge.
Participants using AI made weaker decisions than those using human clinicians.
LLMs perform well on tests yet falter with real patient scenarios.
Details:
Researchers from Oxford tested how people respond to health scenarios using AI chatbots or traditional care. Nearly 1,300 participants reviewed symptoms and chose diagnoses and actions. Results showed chatbots often mixed accurate tips with errors, making it hard for users to spot bad advice. The study argues this gap raises safety risks, including missed emergencies, even though models score highly on medical exams.
Why It Matters:
People already use chatbots as a midnight symptom checker, yet Oxford’s new Nature Medicine user study suggests the weak link is the real chat: with 1,298 participants, people using top LLMs identified the right condition only about 35% of the time and chose the right action about 44%, basically on par with web search and “gut feel”. Add recent safety testing that found a noticeable share of chatbot answers can be problematic or unsafe, and the direction for health products gets pretty concrete: triage flows, red-flag prompts, clearer “get urgent help now” cues, and a fast handoff to a clinician beat open-ended DIY diagnosis.
Better input, better output
Voice-first prompts capture details you forget to type. Wispr Flow turns speech into clean prompts you can paste into your AI tools for faster, more useful results. Try Wispr Flow for AI.
👀 Click on the image you think is real
QUICK HITS
🧠 Deep research in ChatGPT is now powered by GPT-5.2.
💰 Anthropic’s More Than $20 Billion Funding to Close as Soon as Next Week.
📉 US companies accused of ‘AI washing’ in citing artificial intelligence for job losses.
❤️ 'We feel it in our bones': Can a machine ever love you?
🚀 Using AI to help Olympic skaters land a quint.
📈 Trending AI Tools
🗣️ Elevenlabs - Free text-to-speech generator used by many*
🍌 Nano Banana Pro - Google’s AI image generator that’s the best in class
🧠 Skills - Open ecosystem for agent skills
🖥️ Anthropic Cowork - Claude’s Code agent that runs locally on macOS and can read files and take actions
*partner link