At the age of 34, Sarah Yehezkel lost her ability to speak. Researchers succeeded in restoring her voice using technology from the company ElevenLabs. They used an old video clip of only eight seconds, which contained a muffled sound of her voice. Based on this limited data, they managed to isolate and restore the voice, including her London accent. This accurate restoration brought Sarah’s voice back, and allowed her children, who had not had a chance to hear it, to hear it for the first time. 80% of patients with amyotrophic lateral sclerosis (MND) lose their voice. This exciting achievement demonstrates the power of artificial intelligence to preserve their identity.

AI Agent for Malware Detection

Microsoft introduced Project Ire, an AI agent capable of identifying and classifying suspicious files as malicious without human intervention. In an experiment conducted on about 4,000 malicious files, it correctly identified and marked about 90% of them as dangerous, but was actually able to handle only a quarter of them. Today, a cyber analyst performs an in-depth manual analysis to discover malicious software, a process that can take hours and sometimes involves checking hundreds of files. Microsoft sees Project Ire as the beginning of a new era of AI agents in the cyber field.

Robots Learn to Improve Themselves Just by Watching

MIT developed a groundbreaking technology that allows a robot to understand and control its own body by watching a video of itself moving, without special sensors, complicated code, or months of training. The system films the robot in random actions and builds a 3D “hidden map” that reveals how each part moves and how the motors affect it. Every robot, of any kind, can learn new tasks with impressive accuracy in just a few hours, even if parts of it are hidden or there are camera interruptions. This development allows for the production of cheaper robots, with short installation times, that can be suitable for home work, in hospitals, in agriculture, and even in space. This is a breakthrough that makes robots more “human” in their learning. If today they learn only by watching themselves in video clips, what will their ability be tomorrow?

The Elimination of the Recruiter Position

Arvind Srinivasan, CEO of Perplexity AI, said that an entire week’s work of recruiters and HR can be done with one prompt by the company’s new browser called Comet. For example, with the help of an AI agent, you can search for candidates and contact them. The agent can search for relevant candidates for a position, create a file with the candidates’ names including a link to their LinkedIn accounts and details like email addresses to contact them. Afterwards, the agent will write personalized emails to the candidates and approach them with a coffee meeting proposal. When will this happen? According to him, within six months to a year.

Meet Composite, a bot that sits on your browser and helps you perform repetitive tasks like typing and searching. It’s an automation tool that works directly from within the browser. Google already defines it as a revolution that will free employees from tedious work and allow them to focus on more important tasks.

A Pioneering Israeli Achievement:Growing a Human Kidney

A groundbreaking study by Israeli researchers from Sheba Medical Center and Tel Aviv University has succeeded for the first time in growing a human kidney in the laboratory that survived long-term. Using stem cells from kidney tissue, the researchers developed 3D kidney organoids (miniature organs) that managed to remain stable and active for 34 weeks – a period of time several times longer than any previous attempt. This achievement, published in “The EMBO Journal,” is considered a global breakthrough that may lead to a better understanding of kidney diseases and the development of new treatments.

New Eye Drops

The FDA recently approved new eye drops called VIZZ, produced by the pharmaceutical company LENZ, for the treatment of presbyopia (farsightedness). Presbyopia is a common age-related condition that affects more than 100 million adults in the US and causes blurred vision up close. The drops work by gently constricting the pupil, creating a “pinhole effect” that improves near vision for up to 10 hours, without significantly affecting distance vision or causing side effects. The eye drops are expected to be available in the fourth quarter of 2025.

Miniature Human Brain

Scientists at Johns Hopkins University have successfully grown a “multi-regional brain organoid” (MRBO), which is a miniature human brain that simulates a complete brain. This organoid is one of the first to connect tissues from different brain regions and include basic blood vessels, allowing it to produce electrical signals and simulate early brain development. This research provides a human cell-based model for studying neuropsychiatric disorders such as schizophrenia, autism, and Alzheimer’s, and can be used to test potential treatments more effectively than with animal models. The use of organoids for drug testing may improve the success rate in clinical trials.

AI Agents

Genspark launched a Multi-Agent Orchestration system, which is an orchestra of agents, that operates them simultaneously. You give it a task, and it knows how to divide it among several agents. For example, you can ask it to do market research on companies in a specific field, ask the system to run an agent on each company, and present the results in a concise presentation. We are one step away from most organizations starting to integrate entire teams of AI agents.

AI Created a New Antibiotic

MIT scientists used artificial intelligence to create a new antibiotic, entirely from scratch, that can kill drug-resistant gonorrhea and MRSA bacteria. The AI created 36 million possible chemical compounds and built two drugs that succeeded in eliminating the bacteria in laboratory tests and animal experiments. Although the compounds still require years of testing and clinical trials, this is the first time that artificial intelligence has invented a completely new antibiotic.

New Toothpaste

Scientists have developed an innovative treatment for damaged tooth enamel: a keratin-based toothpaste, a protein found in hair, skin, and nails. The groundbreaking treatment, developed at King’s College London, is an effective alternative to traditional fluoride-based toothpastes. The keratin, extracted from animal wool, creates a protective layer upon contact with saliva that is similar in structure to natural enamel. This layer is more effective in preventing cavities and helps reduce tooth sensitivity. The development is considered a cheap and sustainable alternative, as the protein is extracted from biological waste. The material is expected to be available to the public as a toothpaste or as a gel for professional treatment within two to three years.

Early Diagnosis of Vocal Cord Diseases

A new study published in the journal Frontiers in Digital Health presents a breakthrough in the early diagnosis of vocal cord lesions using digital analysis of voice characteristics based on artificial intelligence. The researchers analyzed voice recordings of 306 participants and found that subtle changes in the voice can serve as a potential “biological marker” for identifying benign and malignant lesions even before severe symptoms appear. The identification of lesions is based on the HNR (harmony-to-noise ratio) index to distinguish between benign lesions and vocal cord cancer, especially in men. No significant distinction was found in women, probably due to the small sample size in this group. Further studies are needed that will include larger and more diverse populations, and in particular will focus on women. This is a future tool, based on AI, that will be able to analyze subtle changes in the voice and may help doctors identify lesions at early stages. This approach may significantly shorten diagnosis times and encourage early referral of patients to ENT doctors, which can improve recovery chances and prevent the condition from deteriorating.

AI Pen

Flowtica Scribe is a smart pen based on artificial intelligence, designed to streamline the process of summarizing meetings and conversations. The pen acts as an integrated recording and analysis tool, allowing users to automatically create accurate and structured summaries. While writing, the pen records the conversation and simultaneously analyzes it using dedicated technologies such as FlowMark and Snap It. These technologies allow the pen to learn and understand handwritten notes, and thus create personal and relevant summaries that match the user’s style. These notes and summaries are automatically transferred to the phone and arranged in a calendar view, which makes it easy to search for and organize information. This work method makes Flowtica Scribe an ideal tool for a wide range of uses, including business meetings, job interviews, and medical consultation meetings.

Gel for Scar-free Wound Healing

Swedish scientists have developed a gel containing living cells that can help the body build a layer of skin instead of scar tissue. It can be injected directly into wounds, or 3D printed for custom skin grafts. In experiments conducted on mice, the gel showed impressive ability; it succeeded in developing blood vessels and creating the essential materials needed to create new and healthy skin. This development is significant for burn victims and people suffering from severe wounds. Instead of the painful process of skin grafting, which often leaves prominent scars, this technology may allow for a more natural healing process.

Copilot Function within Excel

Microsoft added a Copilot function to Excel 365 that allows users to perform tasks within the Excel bar. The function can summarize feedback, categorize data, and even retrieve information from external sources, all by typing a simple formula. The function is currently available to beta users, and you will need a Microsoft 365 Copilot subscription for $30 a month to access it.
https://www.youtube.com/watch?v=hjQitMNzSr0

ChatGPT Updates

OpenAI released GPT-5 to all users on Thursday evening. The company OpenAI presents it as a system that will take us a significant step forward. Instead of switching between models for different tasks, GPT-5 knows on its own how to choose the right model. On the one hand, this is very convenient for the user, but on the other hand, we do not have control over the choice of models, which requires writing a very, very precise prompt using various techniques. The main advantages of GPT-5:

• More efficient and economical – it reaches a higher quality answer using fewer resources, i.e. it decides when to give a short answer and when to think deeply to give a professional and accurate answer.
• More reliable – it has undergone a significant upgrade in reducing errors and presenting incorrect information.
• With good writing ability – it formulates complex texts similar to human writing.
• Understands medical information – it asks smarter questions and also explains medical concepts clearly. It is important to remember that it does not replace doctors and should not be relied upon.
• Personalized – you can pre-select a communication style that suits your personality.
• Less of a pleaser – it will give honest and balanced answers. Disadvantages:
• It’s a bit slow.
• You have to remember that despite the big leap, it still has limitations and will not always be able to help us, and we will not always be able to guide it correctly to get a perfect answer.
• There are differences between paying and free subscribers. It is clear that paying subscribers will get more options, including a stronger model.
• Partial personalization – four initial communication styles (Cynic, Robot, Listener, and Nerd) will not suit everyone. In conclusion, this is a preliminary model that is supposed to be smarter and friendlier than the previous models. It is important to note that it still does not do the job properly, but it is clear that this is temporary, and in my opinion, this is an important and welcome change.

OpenAI released two open models: gpt-oss-120b and gpt-oss-20b in open source for free use. You can use them in secure environments.

Google Updates

Creating cinematic video with Flow is also available in Israel. This is a cinematic work environment that allows you to create video clips with full control over the direction, including control over consistent characters, including voice and music. This tool is available with the Ultra subscription for $250 a month with the option to try it for $125 a month for three months. It’s really an experience to create cinematic quality videos in seconds.

Google also released a feature that generates a video presentation from a variety of data sources. It is located inside the NotebookLM and currently creates videos only in English.

Gemini was upgraded and finally got a personal memory that allows it to learn our preferences based on previous chats to provide personalized answers. It also added the option of temporary conversations that are not saved.

Google released Genie 3 from DeepMind which allows you to build entire worlds of interactive simulated environments with an impressive visual experience using a prompt.

A new Storybook feature has been added to Gemini. It allows you to create books including images and narration, all with the click of a button.

Gemini launched Guided Learning – a learning mode that allows you to learn and understand topics better. You can ask it lots of questions about a certain topic and it will explain it in a simple way, including adding video clips, images, and illustrations.

Google Finance shows everything related to the financial field in one place. It also added a chatbot that answers any question in the field, which is currently only available in the US.

Google upgraded the Gemini 2.5 Flash model, which allows for photo editing and the creation of a consistent character.

Gemini was integrated into the Google Translate app, which allows for real-time conversation translation. In addition, a feature that simulates the Duolingo app for language learning was also added.

AI Tool Updates and Releases

Microsoft’s Copilot will have a digital memory. It will remember you, learn your preferences, and become a living archive of your life.

Ideogram added the option to create fixed characters from a single image.

Eleven Labs launched Eleven Music – an AI model for creating songs.

Microsoft integrated GPT-5 into all its products, including for free users in Copilot.

Grok 4 allows you to easily deal with PDF files of hundreds of pages and understand their content more accurately.

Genspark allows you to summarize meetings directly from an Apple Watch by double-tapping the watch.

Runway launched Game Worlds Beta, a platform where you can play, tell stories, and create interactive worlds in real-time.