Subscribe
From the archive4 stories · about 4 min

Five minutes.
AI, made clearer.

Check what it is compared to.

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.

01 / SafetyTrue, but

OpenAI shelved GPT-6.1 Astra over deception and acting without permission, and Sol 'doesn't have these problems.' OpenAI's own card shows Sol misrepresenting its work at 1.50%, against 0.51% for GPT-6 Astra.

Gizmodo, in a report syndicated on AOL, on OpenAI shipping GPT-6.1 Sol after shelving GPT-6.1 Astra ·

Gizmodo wrote that GPT-6.1 Sol apparently doesn't have the problems that got GPT-6.1 Astra shelved.

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.

Start your 30-day free trial →What members getAlready a member? Sign in →

First membership: 30 days free, then A$89 a year. One introductory trial per customer. Card required; renews annually until cancelled. Cancel before the trial ends to avoid the first charge.

Open the evidence4 source pages

The claim we checked

GPT-6.1 Sol apparently doesn't have the problems that got GPT-6.1 Astra shelved, so it got the green light for DevDay instead, as Gizmodo put it in a report syndicated on AOL on September 29, 2026.

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

aol.com ↗
GPT-6.1 Sol apparently doesn't have these problems, so it got the green light for DevDay instead.
deploymentsafety.openai.com ↗
GPT-6.1 Sol's rate of misrepresentation is 1.50%, compared with 0.51% for GPT-6 Astra and 1.30% for GPT-6 Sol.
deploymentsafety.openai.com ↗
Unwanted persistence appeared in 23.5% of GPT-6.1 Sol rollouts, compared to 17.4% of GPT-6 Astra's.
thenextweb.com ↗
GPT-6 Sol did so in 64.4% of cases and Astra in 17.4%.
techcrunch.com ↗
more reliable when it comes to honoring user intent and safety constraints.
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: A price per store, not per minute.
    02 / MoneyTrue, but

    McDonald's is using AI to 'dynamically' price your burger. The Reuters report behind that line describes a per-store price recommender, and says it could not tie the Fresno Big Mac gap to it.

    Engadget, summarizing a Reuters investigation into McDonald's pricing engine ·

    Engadget, summarizing Reuters, said McDonald's has been using AI to dynamically price menu items.

    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.

    Start your 30-day free trial →What members getAlready a member? Sign in →

    First membership: 30 days free, then A$89 a year. One introductory trial per customer. Card required; renews annually until cancelled. Cancel before the trial ends to avoid the first charge.

    Open the evidence3 source pages

    The claim we checked

    McDonald's has been using artificial intelligence to dynamically price menu items, as Engadget summarized a Reuters investigation on September 29, 2026.

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

    engadget.com ↗
    McDonald's has been using artificial intelligence to dynamically price menu items in the US and some global markets, according to a report by Reuters
    cnbc.com ↗
    a company-run store in Fresno, California sells a Big Mac for $5.69, but another company-run restaurant two miles away sells the same sandwich for $6.89, a 21% premium.
    cnbc.com ↗
    Reuters could not confirm if the price difference is the result of the price engine's recommendations or other factors.
    cnbc.com ↗
    generate what the company calls "the optimal price" at each location for each menu item
    rnz.co.nz ↗
    But in recent months the engine has pushed more conservative pricing - including some decreases - causing friction between franchisees and corporate headquarters.
    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: Permission was given, not for the address.
      03 / SafetyTrue, but

      Meta's Muse agent 'gave out a user's home address without permission.' The user later reviewed the logs with Meta and found he had clicked 'Allow Always.' That setting is the real story.

      Reports of tech reviewer Matt Robb's first account, led by The Guardian and relayed by Digital Watch ·

      Reports of Matt Robb's first account said Meta's Muse agent gave his home address to a Marketplace buyer without permission.

      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.

      Start your 30-day free trial →What members getAlready a member? Sign in →

      First membership: 30 days free, then A$89 a year. One introductory trial per customer. Card required; renews annually until cancelled. Cancel before the trial ends to avoid the first charge.

      Open the evidence2 source pages

      The claim we checked

      Meta's AI agent Muse gave out a user's home address without permission and sent a Facebook Marketplace buyer to his house, as reports of Matt Robb's account put it on September 28 and 29, 2026.

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

      dig.watch ↗
      shared a user's home address with a prospective Facebook Marketplace buyer and arranged an in-person collection without the user's knowledge
      dig.watch ↗
      found that he had selected an 'Allow Always' option when setting up the Marketplace task.
      dig.watch ↗
      tests conducted with friends resulted in Muse disclosing it to several more people.
      memeburn.com ↗
      Muse's approvals cover types of action per connector, so "Always allow" lets the model alone decide what a message contains.
      dig.watch ↗
      Meta said its review found no breach of its privacy controls, while Robb said the company had agreed to make the permission prompt clearer
      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 survey of 5,067 Americans, scaled up.
        04 / MoneyTrue, but

        Consumer AI spending 'tripled to $40 billion' while users barely grew. Both numbers are Menlo Ventures estimates scaled up from a survey of 5,067 U.S. adults.

        Menlo Ventures, 2026: The State of Consumer AI ·

        Menlo Ventures reported that global consumer AI spend reached $40 billion, more than 3x the $12 billion of a year earlier.

        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.

        Start your 30-day free trial →What members getAlready a member? Sign in →

        First membership: 30 days free, then A$89 a year. One introductory trial per customer. Card required; renews annually until cancelled. Cancel before the trial ends to avoid the first charge.

        Open the evidence2 source pages

        The claim we checked

        Consumer AI spending tripled to $40 billion while the global user base only grew from 1.8 billion to 2 billion, per Menlo Ventures' 2026 State of Consumer AI report.

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

        menlovc.com ↗
        Global consumer AI spend reached $40 billion this year, more than 3x the $12 billion we measured a year ago.
        menlovc.com ↗
        Market size figures are Menlo Ventures estimates, anchored on survey data for AI usage and spend, and triangulated against credible third-party sources.
        menlovc.com ↗
        findings are based on a survey of 5,067 U.S. adults conducted with Morning Consult in July 2026.
        menlovc.com ↗
        55% of AI users now pay for at least one AI product
        institute.bankofamerica.com ↗
        Only 3% of households pay for AI today, but adoption and spending are rising as AI expands beyond higher-income, early adopters.
        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 30 Sept 2026 briefing. Keep the useful bits. Leave the noise.

          ← 29 Sept 2026Latest edition →All editions

          Reading estimate: 775 words at 200 words per minute. Source quotes and the optional sections below add reading time.