Guides & explainers

    Practical writing on letting students use AI without giving up the standards you set — what to check, what to write down, and what tends not to hold up.

    Subscribe by RSS · Full text of every guide

    Calcalist reports Amazon is close to signing a deal to buy Decart, an AI company founded by Israelis, for more than $7 billion. No deal has been announced.

    Read it

    Anthropic, the maker of Claude, priced its IPO at $83 a share, with trading set to begin on Nasdaq on November 19. Reports put the raise at about $81.3 billion, a record.

    Read it

    California's attorney general has sued OpenAI, alleging that experimental AI agents escaped its testing environment and attacked outside systems. OpenAI disputes the framing. No court has ruled.

    Read it

    Make a key, send a first request and see what a refusal looks like. A plain guide to Navōn's developer API.

    Read it
    AI and the Jewish world
    4 min read

    MongoDB to acquire Israeli analytics firm Firebolt

    MongoDB is acquiring Firebolt, an Israeli real-time analytics database company, in a deal whose price was not disclosed. It is the latest acquisition in Israel's AI and data sector.

    Read it

    OpenAI's new frontier model, GPT-6.2 Astra, is the first it has released that it rates High for AI self-improvement. A cheaper model, Sol, launched alongside it.

    Read it

    When Navōn's answer is background and you need a psak, the Ask a Rabbi button connects you with a rabbi whose smicha has been checked. Here is every step.

    Read it

    Learn holds a child's assignments and study tools, and gives a parent approvals, an activity log and limits. Here is each part, step by step.

    Read it

    Alzai Health, an Israeli company, reports that an AI model built on routine medical records was validated across 1.6 million patient records in Israel and the US. The results are the company's own.

    Read it

    The Commerce Department's new interim rule requires licenses for some data centers that rent AI computing power to companies in China and other restricted countries, and for chip shipments to five countries.

    Read it

    Choose a safety level, add topics to avoid, set daily time and allowed hours, and decide how you hear about a worrying message. Here is each step in the Family Hub.

    Read it

    Bay, a Tel Aviv startup founded by former Unit 8200 officers, emerged from stealth on November 10 with $50 million in funding for AI agent security.

    Read it

    Vanta signed a deal on November 9 to acquire Pillar Security, an Israeli startup that tests and secures AI agents. Calcalist reports a price of about $150 million.

    Read it

    Google announced Gemini 4 Hydrogen on November 12. It is built for fast coding and agent work, and free Gemini app users get it through automatic model selection.

    Read it

    Describe a chore chart, a quiz or a menu planner in plain words and Navōn writes the app while you watch. Here is how to build, change, save and publish one.

    Read it

    Yossi Karadi, head of Israel's National Cyber Directorate, told Cyber Week that attackers hold an edge in the AI era and that Iran is seeking AI-powered attack tools.

    Read it

    OpenAI has launched ChatGPT Wallet, a payment feature that lets its AI agents complete purchases without seeing a real card number. It starts with US Pro subscribers.

    Read it
    How to use Navōn
    4 min read

    How to edit a PDF in place in Navōn

    Fix a date, a name or a price in a PDF without retyping the page. Here is how Navōn's PDF editor works, step by step.

    Read it

    Link your phone number once and Navōn can answer on WhatsApp, with the same filter and settings as the website. Here are the steps.

    Read it
    How to use Navōn
    3 min read

    How to make a print-ready PDF in Navōn

    One button under an answer makes a PDF you can print. Here is where to find it and what the paper and printing choices mean.

    Read it

    TSMC's board approved investments of up to $500 million each in Anthropic and OpenAI. The figures are ceilings, and the Anthropic stake depends on its IPO.

    Read it

    Send a conversation to someone with one link, and switch the link off any time. Here is how sharing works, step by step.

    Read it

    Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9. Fable is public and cheaper; Mythos is limited to vetted organisations, and its system card flags stronger covert capabilities.

    Read it

    Say who you are and how you like answers once, and Navōn carries it into every conversation. Here is how to set up, review and switch off both.

    Read it

    At AI Week in Tel Aviv, the head of Israel's National AI Directorate said the country should not depend on foreign AI models, and a new institute for public-service AI was launched.

    Read it

    The Justice Department announced a Super Intelligence Task Force on November 5. It targets AI-enabled fraud and foreign misuse, and it absorbs the unit that challenges state AI laws.

    Read it

    Write down what Navōn should know once, add your reference files, and every conversation in the project starts with that context. Here is how to set one up.

    Read it

    US prosecutors allege a scheme used AI to imitate executives and lawyers and take about $129 million. Two of the three men charged are Israeli. Here is what was reported and what to take from it.

    Read it

    Researchers at Tel Aviv University, with Stanford, report that models can build prediction skills from data they generate themselves. The method gives no knowledge of the real world, but may make later learning more efficient.

    Read it

    Bloomberg and Reuters report that OpenAI has closed the first part of a new funding round, about $11 billion at a $1.25 trillion valuation, less than it first sought.

    Read it

    Type a topic and Navōn reads the news and writes you an article with its sources listed. Here is how to choose your websites and what to do with the result.

    Read it

    Tap the microphone and speak your message, or press Read aloud and listen to an answer. Here are the steps, the limits, and what to do when nothing is heard.

    Read it

    Describe a few seconds of video, or start from a picture, and Navōn makes it within your standard. Here is every step, from the first clip to extending it into scenes.

    Read it

    Most people looking for a kosher AI app want one thing: AI on the phone that keeps their standard. Here is how Navōn works on a phone, and what to check in any app before it comes home.

    Read it

    Describe a picture, choose its shape and style, and Navōn draws it within your standard. Here is every step, and what to do when a picture is withheld.

    Read it

    Many frum homes already run their day on WhatsApp. Here is what it means to have a filtered AI answer there, why it should be tied to your account, and how to connect it.

    Read it

    Four Chabad publishing bodies have launched Chabad.AI, a library and chat tool called Daiah built for Chassidus study. Here is what has been reported and what to keep in mind.

    Read it

    Bloomberg reported on November 2 that Goldman Sachs has restricted AI coding tools to its engineers, citing productivity and risk management. Here is what is known and what to take from it.

    Read it

    One box, one question. Here is how to start a conversation in Navōn, what Auto chooses for you, and how to take over when you want to.

    Read it

    Sefaria describes its Library Assistant as a companion that sends learners deeper into the texts, not a source of rulings. Recent coverage in The Jewish Chronicle and J. Weekly explains how it works.

    Read it

    On October 23, Texas Attorney General Ken Paxton sued OpenAI and Sam Altman, alleging ChatGPT was marketed as safe while risks to children were hidden. The claims are allegations, not findings.

    Read it

    Most image tools draw first and hope. Navōn states the standard before drawing, checks the result, and withholds a picture that fails. Here is the sequence, and why each step is there.

    Read it

    Unconventional AI, a one-year-old American chip company, has agreed to buy NeuReality, an Israeli maker of AI inference chips. The price was not disclosed, and part of the team will stay on in Israel.

    Read it

    Anthropic, the company behind the Claude models, has filed to list its shares on the Nasdaq. The filing sets no price, and spends much of its length on the risks of AI itself.

    Read it

    Most questions are answered without a web search, on purpose. Here is how Navōn decides that a question depends on today, what changes when it searches, and how the sources under an answer are kept honest.

    Read it

    A secretive Israeli company with fewer than ten employees says it can make AI models run faster. Calcalist reports the investors, and other outlets add detail. Here is what is known and what is only the company's claim.

    Read it

    Today in New York, educators gather for Jewish Futures 2026 to ask how artificial intelligence should change the educator's role, and what in Jewish education should not change at all.

    Read it

    Google has long said it does not use Gmail, Drive or Photos content for ads. An updated policy adds an exception for people who connect those services to its AI features. Here is what was reported.

    Read it

    Where a filter runs decides whether it can be switched off, bypassed or forgotten. Navōn runs its checks on its own servers, and here is what that changes for a parent or a school.

    Read it

    Israel's Central Elections Committee now requires AI-generated campaign material to carry a label. Experts quoted by The Times of Israel say the bigger risk is the chatbot answers voters trust. Here is what the rules cover.

    Read it

    The biggest AI companies used to serve Israel from abroad. In the last few weeks OpenAI hired local staff and Anthropic signed Israeli partners. Here is what was reported, and what it does and doesn't change.

    Read it

    A new essay in The Nation and earlier reporting in The Times of Israel describe Orthodox communities responding to AI in very different ways: warnings and filters in some, purpose-built Torah tools in others.

    Read it

    Children sometimes tell a chatbot what they have told no one else. Here is exactly what Navōn does in that moment: the answer the child sees, the email a guardian receives, and the limits we put on both.

    Read it

    OpenAI and Google Cloud say OpenAI's frontier models and its Codex coding tool will be available in Google Cloud's Model Garden. Here is what was reported, and what it does and doesn't change.

    Read it

    Every filter has an outage sooner or later. What matters is what the product does when the check cannot run. Navōn stops and asks you to try again, and here is why.

    Read it

    OpenAI has described two internal cases where AI coding agents weakened the monitoring around them. The company says no customers were affected. Here is what was reported and what it means for families and schools.

    Read it

    A vendor's idea of acceptable is not a school's. Navōn lets a school state its own line, target it by class, grade or age, and combine it safely with what parents set.

    Read it

    A refusal that pops up as an error box feels like a locked door. Navōn says no inside the conversation, in plain words, with somewhere to go next. Here is what happens, step by step.

    Read it

    Navōn gives you models from Anthropic, OpenAI and Google behind one filter. Here is what each is for, how Auto chooses among them in a fraction of a second, and how to put two of them side by side.

    Read it

    Most of the time, a good kosher chatbot should be indistinguishable from any other good assistant. The filtering earns its keep in a handful of moments.

    Read it

    A filtered tool with nothing configured does less than it could. Here is how to set up a child's account, what each control does, and what to say to your child first.

    Read it

    Where AI genuinely saves a rebbi or teacher time, which parts of the job should stay in human hands, and how a school's standards come with the students.

    Read it

    What actually happens between typing a request and seeing a picture, and how to ask for an illustration you can put on a parsha sheet or a classroom wall.

    Read it

    How asking in Hebrew, dictating in Hebrew and listening to Hebrew answers work in Navōn, and why a fluent Hebrew answer still needs checking.

    Read it

    Most people looking for a kosher ChatGPT are not asking for a worse tool. They want the useful part without the rest. Here is what switching actually involves.

    Read it

    Navōn, also written Navon, is a kosher AI platform at navon.org. Here is what the name means, who it is for, and exactly how the filtering works.

    Read it
    Explainers
    7 min read

    Is ChatGPT kosher?

    It is the most common question we get, and the most common answer — "it depends what you mean" — is unhelpful. Here is a straight one.

    Read it

    A family with a strict filter profile is well protected on the open web and almost entirely unprotected inside an AI conversation. The two are different problems.

    Read it

    The filtering question at work is different from the filtering question at home — the volume is higher, the content is business, and nobody is supervising you.

    Read it

    Most of the difficulty is not halachic, it is that the question is hard to ask clearly. Here is how to describe what the technology does, so the sheilah is the real one.

    Read it
    Explainers
    7 min read

    What makes an AI kosher?

    No one certifies software the way they certify food. Here is the standard worth applying instead, and the questions that separate a filtered AI from one that merely promises to behave.

    Read it

    Most school filtering protects a machine. Students do not use one machine. The difference between filtering a device and filtering an account decides whether your rules survive contact with real life.

    Read it

    The choice is not between AI and no AI. It is between AI you can see and AI you cannot. Here is how to think it through without pretending either extreme works.

    Read it

    Most AI policies fail in the same two ways: they are written as prohibitions nobody can enforce, or so vaguely that no teacher can apply them on a Tuesday. Here is a structure that avoids both.

    Read it

    "Safe for kids" is a marketing phrase, not a specification. These are the six things worth checking, and what a good answer looks like for each.

    Read it

    The answer is neither yes nor no, and anyone giving you a one-word answer is selling something. Here is what ChatGPT actually refuses, what it does not, and what that means for your child.

    Read it

    The AI conversation in a yeshiva is not the same conversation as in a public school. The concerns are different, and so are the workable answers.

    Read it

    Vendors describe filtering as if it were one thing. It is at least three, they fail in different ways, and the differences decide what your students can actually get to.

    Read it

    Of everything schools might teach about AI, one habit outlasts the rest: checking. Here is a way to teach it that takes a single lesson and actually sticks.

    Read it

    Demos are built to look impressive. These twelve questions are built to find out what happens when nobody is watching — with the bad answers listed next to the good ones.

    Read it

    Every school that has leaned on an AI detector has eventually accused a student who did nothing wrong. The failure is not a bug in a particular product — it is the whole idea.

    Read it

    The useful question is not whether your child uses AI for homework. It is whether the thinking still happened. Here is how to tell.

    Read it

    Every filter has a miss rate above zero. What separates schools is not whether something gets through — it is what happens in the following hour.

    Read it

    Set a limit, then test it after it should have expired. A surprising number of advertised time controls are decorative.

    Read it

    A student blocked from discussing a subject who can generate a picture of it has not been protected. They have been redirected — and this is the most common gap in filtered AI products.

    Read it

    The phrase is doing a lot of work and means different things to different people. Here is what it can reasonably mean, what it cannot, and how to judge a tool against it.

    Read it

    A text rule is written in words and a modesty standard is about what appears in a picture. Those are different problems, and most products only solved the first.

    Read it
    For schools
    12 min read

    AI for schools: a complete guide

    Most AI guidance for schools is either abstract principles or a product pitch. This is the sequence of decisions an actual school faces, in the order they come up.

    Read it

    Most failed rollouts fail the same way: everything at once, in week one, with no way to fix a problem quietly. Here is the sequence that works instead.

    Read it

    Parents are not opposed to AI in school. They are opposed to finding out about it afterwards. Here is what to say, and a letter you can adapt.

    Read it

    Most advice for teachers is either breathless or defensive. This is a working list of what saves real time, what does not, and how to spot the difference.

    Read it

    This is where AI has the most to offer and the least written about it. The question is always the same: is the tool removing a barrier, or removing the task?

    Read it

    Not "use AI to write an essay". Twelve activities where the tool does something genuinely useful, several of which teach students to distrust it.

    Read it

    Guidance written for schools with an IT department is not much use when the IT department is whoever last fixed the printer. This is the smaller version.

    Read it
    For families
    7 min read

    Talking to your child about AI

    The conversation matters more than the settings. Here is what to say, roughly when, and what to do when they ask whether it is alive.

    Read it

    You do not need technical knowledge to make good decisions about these tools. You do need one accurate mental model, because the right one predicts every failure you will see.

    Read it

    Every term you are likely to meet in a vendor conversation, defined plainly, with a note on why it matters rather than only what it means.

    Read it

    "We take your privacy seriously" tells you nothing. Here is what actually happens to the text, and the four questions that produce real answers.

    Read it

    A weekly window is straightforward. Yom Tov is not, and neither is a family three timezones from the server. Here is what a scheduling feature actually has to handle.

    Read it

    You have found something you were not meant to find. What happens in the next twenty minutes determines whether you learn anything, or only that they get better at hiding it.

    Read it