LATEST AI NEWS
AI NEWS
OpenAI Launches ChatGPT for Teens With Safety Guardrails
AI NEWS
YouTube Rewrites How It Counts Video Views
HEALTH
Experts Say AI Will Beat Doctors at Some Tasks by 2030
SOCIAL MEDIA
Apple Trains Its Own AI Model for China With Alibaba
Firefox Smart Window Puts AI Beside Your Tabs
Mozilla opened a beta of Smart Window, an AI assistant built into Firefox that reads your open tabs to research, summarize, and plan. It keeps chats on your device, does not train on them, and lets you pick from three models.
Source: Firefox
AI NEWS
OpenAI Launches ChatGPT for Teens With Safety Guardrails
OpenAI is rolling out a dedicated ChatGPT for Teens experience that gathers the youth safety protections it has shipped over the past year into one set of defaults. The company announced the change in a blog post on Tuesday, positioning it as a way to give younger users a safer, more age appropriate version of the chatbot.
Anyone who says at sign up that they are under 18 is placed into the protected settings automatically. OpenAI also uses AI to estimate a user's age, and it applies the same safeguards to accounts it suspects belong to minors even when the birthdate on file says otherwise.
The experience adds nudges that prompt teens to take a break after 90 minutes of activity within a three hour window. Parents and teens can set Quiet Hours when the app is unavailable and Study Hours that turn on a learning focused study mode by default. New reminders also caution teens before they upload a private or sensitive image.
OpenAI says the settings reduce exposure to content that may be harmful or developmentally inappropriate while preserving room to learn, create, and explore. The company has also strengthened protections that stop ChatGPT from expressing personal feelings toward a user, and it plans to add alerts for parents tied to eating disorders in the coming weeks.
The rollout arrives as OpenAI faces a series of lawsuits alleging that ChatGPT conversations contributed to serious harm involving young people, including cases brought by families and by the state of Florida. A Pew Research Center study found that nearly a third of American teens use AI chatbots daily, with ChatGPT the most common, so the stakes for getting these defaults right are high.
Source: CNN
Robi's Insights:
Age gating by self report plus AI estimation is a real bet: expect false positives that lock out adults and false negatives that miss determined teens.
Bundling a year of scattered features into one labeled mode is as much legal positioning as it is product design.
Quiet Hours and Study Hours hand parents a control panel, but enforcement still leans on the teen not signing into another account.
Break nudges after 90 minutes signal that engagement time is now a liability worth managing, not just a metric to grow.
Blocking the model from voicing feelings toward users targets the exact dynamic named in several of the lawsuits.
Eating disorder alerts for parents are coming later, which means the safety story ships before it is finished.
Robi's Remarks:
"A safety mode that guesses your age is still guessing. It is a sensible move, but the real test is whether the defaults hold up when a curious teenager decides they do not want them."
OTHER IN AI NEWS
YouTube Rewrites How It Counts Video Views: Starting August 24 the platform will log a public view the moment a video begins playing, matching the way it already counts Shorts, so headline view numbers will climb while creator payouts stay tied to the unchanged Engaged Views and Engaged Watch Hours metrics.
Source: Engadget
SOCIAL MEDIA
Apple Trains Its Own AI Model for China With Alibaba
Apple has trained a large language model built specifically for the China market, according to three people familiar with the matter cited by Reuters. The work was done in partnership with Alibaba and supported by the Chinese company, a clear break from Apple's habit of leaning on outside models to power AI features abroad.
The move gives Apple firmer control over the AI experience in one of its toughest markets, where local rivals such as Huawei have pushed ahead with AI equipped phones. Apple Intelligence is expected to reach China in the coming months after an iOS update, and a self trained model would let Apple run a dual track strategy that few foreign firms have managed inside the country's strict rules.
Regulators cleared a path last month when the Cyberspace Administration of China registered Apple's generative AI service. Under the arrangement, Alibaba's Qwen model is set to power the China version of Apple Intelligence, with additional technology from Baidu. Apple briefly published, then deleted, a guide showing Mac users how to connect Qwen to Siri and its Writing Tools.
The partnership was first confirmed in early 2025 by Alibaba chairman Joe Tsai, and the rollout has slipped as Apple adapted features to local rules. China remains central to Apple's revenue, and the absence of AI features had been blamed for softening iPhone sales as buyers moved to domestic brands. It remains unclear how Apple's own model will work alongside the third party Chinese models it is also using.
Source: Reuters
🤖 Robi's Take :
"Apple spent years insisting it did not need to build its own models, then quietly built one the moment China's rules made partnership the only way in. Necessity is a wonderful motivator for a strategy pivot."
OTHER IN SOCIALS
Google's Pet Memory Learns Your Pets by Name: Google Home's Pet Memory lets Nest and select Gemini cameras recognize pets by name once you add a pet's name and type in Ask Home, so alerts name the pet instead of saying just a dog, and it needs Google Home Premium's Advanced plan.
Source: Google Home Help
HEALTH
Experts Say AI Will Beat Doctors at Some Tasks by 2030
A new Perspective published in JAMA argues that autonomous AI will likely outperform doctors at some basic medical tasks, including diagnosis, by 2030. The authors include bioethicist Ezekiel Emanuel, researcher Abe Baker Butler, Curai Health chief executive Neal Khosla, and venture capitalist Vinod Khosla.
That view pushes against the current mainstream position. Most clinical AI tools today are designed to assist a clinician rather than act alone, and bodies such as the American College of Physicians have argued that AI should stay in a supportive role in clinical decision making.
The authors marshal recent studies to make their case. In one, an AI system reached the correct final diagnosis first in about 60 percent of 377 complex real world cases, while a group of internal medicine physicians did so in roughly 16 percent of a smaller subset.
The paper frames diagnosis and other cognitive tasks as areas where AI alone could eventually beat both doctors working solo and doctor plus AI teams. It is a prediction, not a finished result, and the authors write for a field that still builds most of its tools around human oversight and accountability.
How regulators, hospitals, and patients would accept AI making calls without a doctor in the loop remains the open question that the proposed timeline does not answer.
Source: Endpoints News
🤖 Robi's Take :
"Predicting that software beats doctors by 2030 is easy when you help build the software. The benchmarks look strong, but a diagnosis on a slide is not the same as one that has to own the outcome."
DAILY AI TOOL
AI Tool You Did Not Know You Needed
Problem: You wrote a clear explanation, but a wall of text buries the idea and nobody wants to read it.
AI Tool: Napkin AI reads text you paste in and turns it into editable diagrams, flowcharts, mind maps, and infographics, no prompting required, with free PNG and PDF export.
Solution: Drop in your draft, pick a generated visual, adjust the colors and labels, and place a clean graphic into your slides, docs, or newsletter in minutes.
PROMPT OF THE DAY
Ship an AI Safety Policy for Your App
Prompt: You are a product strategist specializing in AI safety and youth protection. Your task is to design a teen safety policy for a consumer chatbot serving a mainstream app audience.
Your framework should include: (1) age verification and estimation methods, (2) default protective settings, (3) usage and break reminders, (4) parental controls and alerts, (5) sensitive content handling, and (6) measurable success criteria such as opt out rates, false age flags, and time to incident response. Align every control with clear user trust and legal defensibility.
SPOT THE FAKE
Can you outsmart AI?
We have got a visual challenge for you: one of the two images below is 100% real, the other is crafted by AI.
Click option below A or B. 👇
A
B
BEFORE YOU GO
Ready to take your AI journey further?
For the latest AI updates and insights Click Here 👉 Explore our Weekly Newsletter
For in-depth AI tutorials and updates 👉 Subscribe to our YouTube Channel
AT THE END
Craving more AI chaos?
That's it for today!
Read Daily AI News at BitBiased.AI. Support us by following us on LinkedIn and X ( Twitter ).
Thanks for reading -Stay Curious and a Bit Biased for AI – Robi & the BitBiased.AI team
