While You Sleep

Scientists at Stanford University have developed a first-of-its-kind AI model called SleepFM, which can predict the risk of over 100 health conditions based on just one night of sleep data. SleepFM was trained on physiological sleep data broken into five-second segments to learn the “language” of connections between body systems during rest.

The model was trained on nearly 600,000 hours of sleep data from 65,000 participants. Researchers linked sleep data to medical records spanning up to 25 years, showing high accuracy in predicting serious diseases years before onset:

The World’s Best Robot

The humanoid robot known as Atlas has won the award for the most advanced and impressive robot. It captivated global media with its stable, natural gait and its ability to perform movements that exceed human range, such as 360-degree joint rotation. Hyundai, which acquired Boston Dynamics, announced an ambitious goal: the production of 30,000 units per year.

A Robot to Strengthen Family Bonds

The company OLLOBOT has introduced an interesting approach to developing social robots. The robot is capable of decoding moods—identifying stress, sadness, or joy in a room—and responding accordingly with sounds and movements reminiscent of a pet. The goal is to spark empathy and curiosity, especially among children. One unique feature is its ability to track extended family networks; if it notices you haven’t spoken to your grandmother or a distant cousin in a while, it will remind you to get in touch.

Sexual Exploitation and AI

Grok users are generating nude and sexual images of famous women, anonymous individuals, and most distressingly, young girls. This model currently operates with zero filters or oversight. Since the acquisition of Twitter (now X), Elon Musk has dismissed most of the Trust & Safety team responsible for monitoring offensive content. An ideology of “absolute freedom” has turned Grok into an ideal tool for bad actors to create harmful content, leading to what many call an AI “Wild West” where authorities often shift responsibility back to the users.

Allergen Testing Device

The French startup Allergen Alert unveiled a compact, book-sized device at CES 2026 that allows for the real-time, lab-grade detection of allergens such as gluten, dairy, and peanuts. Set to launch later this year, the device aims to provide peace of mind and safety for people with severe allergies when dining out.

AI in the Medical Field

Every week, hundreds of thousands of people consult AI regarding medical issues. It often makes fewer errors than doctors and explains complex terms in simple language. This trend led OpenAI to launch ChatGPT Health for personal medical file management and data analysis.

• Anthropic launched Claude for Healthcare, focusing on maximum safety, reducing “hallucinations,” and medical ethics.

• Google released MedGemma 1.5, an open-source model capable of analyzing complex 3D scans.

• Amazon launched One Medical, a 24/7 virtual health assistant familiar with your medical history.

Who is Liable When AI Makes a Medical Mistake?

Last year in Shanghai, four senior doctors competed against AI systems in diagnosing complex gastrointestinal cases. While the AI provided answers in 2 seconds compared to the doctors’ 13 minutes, the event sparked a major ethical debate: who is responsible when AI fails? Currently, the responsibility sits in a gray area between developers, the doctors using the tool, and the healthcare systems adopting it. Clear regulation is needed to ensure transparency and human oversight.

ChatGPT Updates

• ChatGPT Go: A new $8 subscription offering 10x more messages than the free version, fewer upload restrictions, and an upgraded context window (includes ads).

• ChatGPT Translate: A dedicated tool supporting 50+ languages that identifies slang and cultural nuances rather than just literal translation.

• ChatGPT Jobs: An AI career agent that optimizes resumes, provides competitive advice, and actively assists in job searches.

• Prism: A free workspace designed for scientific research and writing, allowing real-time collaboration and direct integration of academic literature.

Google Updates

• Personal Intelligence: Personalized AI providing recommendations based on your data across all Google apps.

• Translate Gemma: Three models with advanced translation capabilities.

• Agent Skills: A new feature in Antigravity allowing developers to use reusable “Skills.”

• AI Inbox: An agent designed to help users reach “Inbox Zero” by focusing on high-priority emails.

• Gemini for Google TV: Enables voice-activated commands for television.

• AI Detection: A new feature for the education system to identify if images or videos were created/edited by Google’s AI.

• Google AI Plus: A discounted subscription (approx. 30 NIS/month) including 200GB of storage.

Microsoft Updates

• Copilot Checkout: Allows users to shop and purchase products directly within the app.

• GitHub Copilot SDK: Now accesses organizational knowledge and meeting transcripts to provide better coding insights.

Claude Updates

• Claude Cowork: An AI agent that performs complex tasks on your computer autonomously without requiring technical knowledge.

• Office Integration: Claude can now edit, analyze, and create Word and Excel documents.

• Claude in Excel: A specialized feature for financial modeling, formula correction, and data scenario running (available for paid subscribers).

• External Integration: Full interaction with tools like Figma, Canva, Monday, Slack, and Box.

AI Tool Updates and Releases

• Amazon revealed Alexa+, a browser-based chatbot focused on family needs and smart home control.

• CES 2026 featured advanced health sensors and home robots designed to improve quality of life.

• Gen & Intel unveiled AI technology that detects voice and visual deepfakes directly on-device for real-time fraud protection.

• Freepik launched Variations, a tool that turns a single image into a storyboard sequence. Runway released a similar tool called Story Panels.

• ElevenLabs released Scribe v2 for high-precision transcription.

• LTX announced the ability to generate video from sound.

• Adobe introduced features that turn documents into interactive content, such as podcasts and designed presentations.

• Wix developed Wix Harmony for building entire websites and pages using AI.

Would you like me to adjust the tone of any specific section to be more formal or more conversational?