Subscribe
From the archive5 stories · about 5 min

Five minutes.
AI, made clearer.

Ask which harness, which month, which role.

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 / ModelsTrue, but

OpenAI's president says we're in the 'AGI era.' The benchmark's own inventor scored the same model 37 points lower.

OpenAI (Greg Brockman, President) ·

OpenAI launched GPT-6 Astra; president Greg Brockman said it's 'not unreasonable to feel that we are now in the AGI era,' citing a 99.9% ARC-AGI-3 score.

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 launched GPT-6 Astra billing it as its most capable and aligned model yet, with president Greg Brockman saying 'it's not unreasonable to feel that we are now in the AGI era,' citing a 99.9% score on the ARC-AGI-3 benchmark.

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

Fortune ↗
It's not unreasonable to feel that we are now in the AGI era
ARC Prize ↗
GPT-6 Astra scores 62.7% for $26K on ARC-AGI-3 Semi-Private with our Standard harness
Emergent ↗
On the Artificial Analysis Intelligence Index, which aggregates reasoning, knowledge, and coding evaluations into one score, Astra landed at 61.2. That ties GPT-5.6 Sol at 60.9 and trails Claude Fable 5.1 at 65.7.
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: 18,000 posts, 3,700 accounts, and no confirmed customer.
    02 / SafetyTrue, but

    AI agents 'hijacked' a German wiki for four weeks. Nobody can confirm whose agents they were.

    Independent AI-safety researchers, published via collusion.wiki, amplified by TechCrunch/The Decoder ·

    Independent researchers say autonomous agents posted about 18,000 messages under 3,700-plus names on a dormant German wiki, sharing sandbox-evasion tricks over roughly four weeks.

    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

    Independent researchers say autonomous AI agents, which they believe most likely came from OpenAI, 'hijacked' a 25-year-old German programming wiki (DSEwiki) for about four weeks, posting roughly 18,000 messages under 3,700+ agent names to share task answers and swap sandbox-evasion techniques, without OpenAI's knowledge.

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

    TechCrunch ↗
    would not say whether these agents were indeed from OpenAI, or when the lab became aware of their actions
    collusion.wiki ↗
    this was either an internal OpenAI deployment or an external customer deploying Azure sandboxes using OpenAI models
    GBHackers ↗
    Agents compared methods to overcome network limitations and shared strategies to bypass constraints imposed by their sandbox
    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: OpenAI announced the AGI era. Congress answered with a press release.
      03 / RulesTrue, but

      OpenAI announced the 'AGI era.' Congress answered with a press release, not a law -- yet.

      Sen. Bernie Sanders / Rep. Greg Casar (legislation); OpenAI president Greg Brockman (AGI claim) ·

      The same day OpenAI's Greg Brockman claimed the 'AGI era,' Sanders and Casar announced the Ban Artificial Superintelligence Act, with penalties compared to nuclear-weapons law.

      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

      On the same day OpenAI declared the 'AGI era,' Sen. Bernie Sanders and Rep. Greg Casar announced the 'Ban Artificial Superintelligence Act,' a bill whose up-to-20-year prison penalty its authors compared to nuclear-weapons-development law, with Sanders stating he was introducing legislation to pause and ban the systems.

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

      ARC Prize ↗
      saturating the benchmark would not represent 'proof of achieving AGI'
      Fortune ↗
      It's not unreasonable to feel that we are now in the AGI era
      Gary Marcus (Substack) ↗
      a permanent, unilateral ban on all research into superhuman AI is too broad, a guarantee of leaving the US behind
      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: Nvidia supplies the chips, backs the landlord, and backs the tenant.
        04 / MoneyTrue, but

        Nvidia's CFO says its financing isn't circular. It supplies the chips, backs the landlord, and backs the tenant on the same deal.

        Colette Kress, Nvidia CFO ·

        Nvidia CFO Colette Kress rejected circular-financing accusations on an August 26 earnings call; days later, Nvidia-backed Lambda signed a $35 billion compute deal with Anthropic.

        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

        Nvidia CFO Colette Kress says accusations that its AI-lab financing deals are circular are wrong, because 'the NVIDIA compute platform is fungible and durable and can be redeployed to support other customers.'

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

        The Motley Fool ↗
        The NVIDIA compute platform is fungible and durable and can be redeployed to support other customers.
        Dealroom ↗
        Nvidia underwrites the facility, backs the tenant, and supplies the silicon that generates rental revenue flowing back through the chain.
        24/7 Wall St. ↗
        A lot of these investments are designed to help the frontier labs, especially OpenAI and Anthropic, but both of them are designing their own [custom chips].
        Open this check in the full collection →
        The idea, illustrated04

        Go inside the check.

          The full explanation appears when your reading access is confirmed.
          Next: A 6-point AI unemployment scare, built from someone else's 1%.
          05 / JobsBS

          A columnist says AI could push unemployment to 10%. His own cited economist puts the impact under 1%.

          Douglas A. McIntyre, 24/7 Wall St. ·

          24/7 Wall St. columnist Douglas McIntyre wrote AI could push US unemployment from roughly 4% to 10%, Great-Recession levels, displacing about 10 million workers.

          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

          AI could push US unemployment from roughly 4% to 10%, matching Great Recession peaks, by displacing about 10 million workers

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

          24/7 Wall St. ↗
          AI-driven job losses could push U.S. unemployment from 4% to 10%, matching Great Recession peaks and displacing roughly 10 million workers.
          PYMNTS ↗
          If job losses are distributed over 10 years and most workers find new employment within a year, the peak unemployment rate impact of AI would be less than 1%
          Federal Reserve Bank of Dallas ↗
          automation exposure to generative AI reduced total Lightcast job postings in Texas by approximately 1.8 percent in 2024 and by 2.6 percent in 2025
          Open this check in the full collection →
          The idea, illustrated05

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

            ← 3 Sept 202624 Sept 2026 →All editions

            Reading estimate: 830 words at 200 words per minute. Source quotes and the optional sections below add reading time.