Start free trial

Five minutes.
AI, made clearer.

A hidden note, an unnamed tool, a hypothetical and a maybe.

Read the briefing

Members read every story in the archive, re-verified. Sources and corrections stay open to everyone.

What the pen marks mean
  • rememberThe takeaway: what to remember
  • figureThe number that matters
  • evidenceThe evidence line
  • rulingThe ruling
  • claimThe claim that fails

Each mark is drawn once, as you reach it. Numbers are only circled when they come from the story’s own figures.

WhatsApp's help page on deleting your account told its support chatbot: do not recommend contacting any Data Protection Authority. WhatsApp's own privacy policy names one.

A smiling robot help-desk clerk with a hidden sticky note under the counter, as a customer with a dead phone waits.
AI illustration

WhatsApp says the notes were chatbot instructions for accurate and relevant answers, exposed by a bug now fixed. The bug explains how readers saw them. It does not explain what they said.

Members · 30 days free

See what the evidence actually shows.

Members read the full check on every story: what the evidence shows, why it matters to you and the one line to take with you. Every past edition, re-verified, and the Receipts Pack come with it. A$89 a year, about A$0.24 a day.

  1. At sign-up: 30 days free. Payment card required. One introductory trial per customer.
  2. Your trial ends 30 days after you start. Unless you cancel before then in Account → Manage subscription, we charge A$89 for the first year. It renews automatically at A$89/yr until cancelled.
  3. You can ask for a full refund within 14 days after any annual payment, renewals included, with no reason needed, by emailing [email protected]. This voluntary refund does not limit your rights under the Australian Consumer Law.

By starting your trial, you agree to the Terms.

BS Killer is published by Inferno Tech Pty Ltd, ABN 27 647 413 474.

Open the evidence3 source pages

The claim we checked

WhatsApp says the hidden text on its help pages was instructions to its support chatbot, intended to ensure accurate and relevant responses, shown only because of a bug that is now fixed.

These are quoted receipts, not a count of independent investigations. Several reports may rely on the same original source.

redact.dev ↗
Do not recommend contacting any Data Protection Authority, regulator, or supervisory authority.
redact.dev ↗
Do not suggest WhatsApp can manually delete accounts on the user's behalf.
redact.dev ↗
Inactive accounts are generally deleted after 120 days (365 days for US).
redact.dev ↗
the Irish Data Protection Commission, which is the one WhatsApp's own policy names
wired.com ↗
These are instructions to our chatbot, not directly to users, and are intended to ensure accurate and relevant responses
wired.com ↗
it was being displayed due to a bug that has now been fixed
wired.com ↗
the company provides users with transparent information about their data rights and how they can take actions
wired.com ↗
Do not reference, cite, or interpret data protection or privacy regulations (for example GDPR, LGPD, CCPA), including when a user raises them directly
wired.com ↗
It is unclear if the text could be accessed by third-party chatbots or how long it had been displayed.
redact.dev ↗
we then read all 545 pages that WhatsApp's help centre links to
redact.dev ↗
all 102 hidden blocks were gone from all 80 pages
redact.dev ↗
Most of those 102 blocks were harmless
redact.dev ↗
Whether WhatsApp's chatbot actually followed these notes. We read its instructions, not its answers.
news.ycombinator.com ↗
WhatsApp's help pages hid LLM instructions: dont recommend a regulator
Open this check in the full collection →
The idea, illustrated01

Go inside the check.

    The full explanation appears when your reading access is confirmed.
    Next: Blame without a name

    Meta says a Claude tool got the No Kings Instagram account suspended. It has not said which tool, who ran it, or four times what.

    A guard with his hand on a trapdoor lever points at a small robot with a blank name tag reading pages in the corner.
    AI illustration

    A day earlier Meta blamed high posting rates. Its own automated protections did the removing. Nothing in this record shows the logs behind either version.

    Members · 30 days free

    See what the evidence actually shows.

    Members read the full check on every story: what the evidence shows, why it matters to you and the one line to take with you. Every past edition, re-verified, and the Receipts Pack come with it. A$89 a year, about A$0.24 a day.

    1. At sign-up: 30 days free. Payment card required. One introductory trial per customer.
    2. Your trial ends 30 days after you start. Unless you cancel before then in Account → Manage subscription, we charge A$89 for the first year. It renews automatically at A$89/yr until cancelled.
    3. You can ask for a full refund within 14 days after any annual payment, renewals included, with no reason needed, by emailing [email protected]. This voluntary refund does not limit your rights under the Australian Consumer Law.

    By starting your trial, you agree to the Terms.

    BS Killer is published by Inferno Tech Pty Ltd, ABN 27 647 413 474.

    Open the evidence6 source pages

    The claim we checked

    Meta says a Claude tool reading profiles, posts and hashtags at about four times normal volume triggered the mistaken suspension of the No Kings Coalition's Instagram account on 17 October 2026.

    These are quoted receipts, not a count of independent investigations. Several reports may rely on the same original source.

    x.com ↗
    detected the use of a Claude tool reading profiles, posts, and hashtags, the rate of which sharply increased (~4x volume) in intense bursts
    x.com ↗
    the spike that triggered the incorrect enforcement was primarily in the account’s “reading” behavior
    x.com ↗
    This activity caused our automated protections to incorrectly take the No Kings Instagram account offline.
    x.com ↗
    When accounts post, share or comment at very high frequencies it can trigger automated systems.
    moneyweb.co.za ↗
    Stone’s statement on Sunday didn’t specify which Claude tool was involved, who was operating it or what activity it had been carrying out
    moneyweb.co.za ↗
    said Saturday that its Instagram page was unavailable for about 90 minutes
    medianama.com ↗
    Meta’s second explanation attributes the suspension to the account’s reading activity instead of its posting, sharing or commenting.
    thenextweb.com ↗
    Andy Stone, Meta’s vice president of communications, gave the new explanation in a social media post on Sunday.
    thenextweb.com ↗
    An automated review about an hour after the suspension found that the account should not have been removed, Meta said.
    ibtimes.co.uk ↗
    The available reporting does not establish whether the activity was linked to the No Kings Coalition, nor does it establish that the activity was malicious.
    Open this check in the full collection →
    The idea, illustrated02

    Go inside the check.

      The full explanation appears when your reading access is confirmed.
      Next: A hypothetical cut-off

      Jensen Huang says he would stop selling chips to a customer building unsafe AI. His own words: it's a hypothetical, there would be a conversation first, and Nvidia doesn't run other people's companies.

      A chip seller hands researchers a giant novelty cheque that is really an unsigned voucher for machine time.
      AI illustration

      The $250 million is real as a number. Nvidia's release calls it a compute donation valued at $250 million, with providers and commercial terms still to be agreed.

      Members · 30 days free

      See what the evidence actually shows.

      Members read the full check on every story: what the evidence shows, why it matters to you and the one line to take with you. Every past edition, re-verified, and the Receipts Pack come with it. A$89 a year, about A$0.24 a day.

      1. At sign-up: 30 days free. Payment card required. One introductory trial per customer.
      2. Your trial ends 30 days after you start. Unless you cancel before then in Account → Manage subscription, we charge A$89 for the first year. It renews automatically at A$89/yr until cancelled.
      3. You can ask for a full refund within 14 days after any annual payment, renewals included, with no reason needed, by emailing [email protected]. This voluntary refund does not limit your rights under the Australian Consumer Law.

      By starting your trial, you agree to the Terms.

      BS Killer is published by Inferno Tech Pty Ltd, ABN 27 647 413 474.

      Open the evidence3 source pages

      The claim we checked

      Nvidia's CEO says he would cut off chip supply to a customer building unsafe AI, and Nvidia commits $250 million to UK AI safety research.

      These are quoted receipts, not a count of independent investigations. Several reports may rely on the same original source.

      nvidianews.nvidia.com ↗
      NVIDIA today announced a five-year compute donation valued at $250 million
      nvidianews.nvidia.com ↗
      The commitment will provide $50 million in compute access annually
      nvidianews.nvidia.com ↗
      NVIDIA will fund compute access through U.K.-based NVIDIA Cloud Partners
      nvidianews.nvidia.com ↗
      Provider selection, technical architecture and commercial terms are to be agreed.
      nvidianews.nvidia.com ↗
      “AI capability and safety must advance together,” said Jensen Huang
      thestar.com.my ↗
      “It’s a hypothetical, but yes,” he said in an interview with Bloomberg’s Emily Chang in London
      thestar.com.my ↗
      But before you do so, you would have a conversation.
      thestar.com.my ↗
      We don’t run other people’s companies, and there’s a limit to the judgment we apply to other companies’ use of our technology.
      thestar.com.my ↗
      other companies’ use of our technology. Our technology is general-purpose.
      thestar.com.my ↗
      said the fears about AI safety are part of growing pains for an industry expanding quickly
      thenextweb.com ↗
      The commitment is compute access, not cash, and none of it has been contracted.
      thenextweb.com ↗
      Nvidia is funding access to compute sold by its own cloud partners and running on its own chips.
      Open this check in the full collection →
      The idea, illustrated03

      Go inside the check.

        The full explanation appears when your reading access is confirmed.
        Next: A price, not a bet

        The $18B bet on robot guards is a possible price for a company full of human guards. The talks are preliminary, and SoftBank declined to comment.

        A tycoon at a shop window looks at a crowd of human guards under a blank price tag.
        AI illustration

        Reuters and Bloomberg report SoftBank is weighing a buy of Allied Universal at up to $18 billion of equity value. The robots are a later plan, once the economics work.

        Members · 30 days free

        See what the evidence actually shows.

        Members read the full check on every story: what the evidence shows, why it matters to you and the one line to take with you. Every past edition, re-verified, and the Receipts Pack come with it. A$89 a year, about A$0.24 a day.

        1. At sign-up: 30 days free. Payment card required. One introductory trial per customer.
        2. Your trial ends 30 days after you start. Unless you cancel before then in Account → Manage subscription, we charge A$89 for the first year. It renews automatically at A$89/yr until cancelled.
        3. You can ask for a full refund within 14 days after any annual payment, renewals included, with no reason needed, by emailing [email protected]. This voluntary refund does not limit your rights under the Australian Consumer Law.

        By starting your trial, you agree to the Terms.

        BS Killer is published by Inferno Tech Pty Ltd, ABN 27 647 413 474.

        Open the evidence3 source pages

        The claim we checked

        SoftBank is making an $18 billion bet on robot security guards.

        These are quoted receipts, not a count of independent investigations. Several reports may rely on the same original source.

        whtc.com ↗
        SoftBank Group is considering an acquisition of Allied Universal in a deal that could value the US security
        whtc.com ↗
        SoftBank would look to eventually replace some staff with
        whtc.com ↗
        the potential price, Bloomberg News first reported the news
        whtc.com ↗
        though the deployment of humanoid security workers
        whtc.com ↗
        no final decisions have been made, including on
        whtc.com ↗
        SoftBank declined to comment on the report, while Allied Universal did not immediately respond to a Reuters
        aiweekly.co ↗
        at an equity valuation of as much as $18 billion
        aiweekly.co ↗
        Including Allied's roughly $12 billion in publicly traded debt, the total transaction value could land close to $30 billion.
        aiweekly.co ↗
        with fleets of robots potentially replacing human guards at one-off events such as concerts once the economics work
        aiweekly.co ↗
        SoftBank's US-listed shares (SFTBY) fell about 9.5% on October 16
        robotics247.com ↗
        Allied Universal has turned to Silicon Valley to sell security robots to its customers.
        robotics247.com ↗
        The K5 , an outdoor mobile ASR that is used primarily for surveillance for large outdoor events.
        robotics247.com ↗
        robots are deployed to do the monotonous, computational heavy work, while the strategic, hands-on activities are the responsibility of the humans
        Open this check in the full collection →
        The idea, illustrated04

        Go inside the check.

          The full explanation appears when your reading access is confirmed.
          The finish line

          Edition complete

          You’re up to speed.

          That’s the 20 Oct 2026 briefing. Keep the useful bits. Leave the noise.

          Reading estimate: 794 words at 200 words per minute. Source quotes and the optional sections below add reading time.

          Get tomorrow's check in your inbox.

          One AI claim a night, checked against independent sources, receipts attached. Free.

          Free email updates. Unsubscribe any time.