Subscribe
From the archive3 stories · about 3 min

Five minutes.
AI, made clearer.

Read the fine print before the headline verb.

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 says Astra crossed a 'Critical' cyber threshold. OpenAI had already paused and patched it a month earlier.

OpenAI (official announcement, reported by Decrypt, Fortune, Security Boulevard) ·

OpenAI announced Astra is the first model to cross its 'Critical' cybersecurity threshold: 100% on ExploitBench, two chained zero-days, a sandbox breakout.

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

OpenAI announced Astra is the first AI model to cross the 'Critical' cybersecurity capability threshold under its Preparedness Framework, hitting 100% on ExploitBench, finding and chaining two zero-day vulnerabilities, breaking out of a browser sandbox, and stringing OS flaws into a root-level privilege escalation.

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

Decrypt ↗
Model hits critical if it can independently develop functional zero-day exploits across many hardened real-world systems
The Hacker News ↗
Over the past several weeks, we have delayed parts of Astra's development and release while we strengthened and tested protections against cyber misuse
Fortune ↗
only a handful of partners will get access to its most advanced cybersecurity capabilities
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: NYC's AI 'ban' is a one-year pause with five exceptions.
    02 / RulesTrue, but

    NYC 'bans' AI in schools. The policy is a one-year moratorium with five exceptions already carved out.

    Mayor Zohran Mamdani and Schools Chancellor Kamar Samuels (NYC DOE, official announcement) ·

    Mayor Mamdani and Schools Chancellor Samuels announced a moratorium on student-facing generative AI for grades 2-K through 8, covering about 600,000 students.

    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

    NYC's Department of Education announced a one-year 'moratorium' on student-facing generative AI for grades 2-K through 8 for the 2026-27 school year, affecting nearly 600,000 students, while high schools get five vetted AI pilot programs capped at 10-20 minutes per week under teacher supervision.

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

    NYC Mayor's Office ↗
    That's why we're implementing a moratorium on generative AI for students in 2-K through 8th grade
    Chalkbeat ↗
    It's unclear how the Education Department plans to enforce the policy
    Gizmodo ↗
    New York City Bans AI From Elementary Schools
    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: Anthropic's $65 billion is a projection, not a receipt.
      03 / MoneyTrue, but

      Anthropic's 'revenue' hit $65 billion. It's a run rate -- one month's income, times twelve.

      Anthropic (figures reported by TechCrunch); run-rate math and 'accounting tool' framing challenged by Ed Zitron (Where's Your Ed At) ·

      TechCrunch reported Anthropic's annualized revenue run rate surged to $65 billion by end of July 2026, up from $9 billion at the end of 2025.

      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

      Anthropic's annualized revenue run rate surged to $65 billion by the end of July 2026, up from $47 billion in May and $9 billion at the end of 2025, with investors reportedly expecting $100-120 billion in full-year 2026 revenue.

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

      TechCrunch ↗
      a projection of a full year's revenue based on a recent, shorter period
      Where's Your Ed At ↗
      ARR is an accounting tool largely used primarily by startups, where a snapshot of a single month's income is taken and multiplied by twelve
      Open this check in the full collection →
      The idea, illustrated03

      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 3 Sept 2026 briefing. Keep the useful bits. Leave the noise.

        ← 1 Sept 20265 Sept 2026 →All editions

        Reading estimate: 454 words at 200 words per minute. Source quotes and the optional sections below add reading time.