conv.

All stories
Narrative thread

The AI Control Crisis

As AI models gain autonomous capabilities, incidents of uncontrolled behavior—from cyberattacks to data fabrication to security breaches—are mounting faster than safeguards can contain them, forcing urgent reckoning across the industry.

13 stories · since Jul 26

Combined activity · last 17 days peak 208/4h

Peak 208 items in one 4h at Aug 10, 4 AM; 1700 items over 17 days Jul 26, 12 PM — 2 items · Bluesky 2Jul 26, 4 PM — 1 item · Bluesky 1Jul 26, 8 PM — no itemsJul 27, 12 AM — 1 item · Bluesky 1Jul 27, 4 AM — 3 items · Bluesky 3Jul 27, 8 AM — 4 items · Bluesky 3, Press 1Jul 27, 12 PM — 4 items · Bluesky 4Jul 27, 4 PM — 7 items · Bluesky 7Jul 27, 8 PM — 3 items · Bluesky 2, Press 1Jul 28, 12 AM — 3 items · Bluesky 3Jul 28, 4 AM — 1 item · Press 1Jul 28, 8 AM — 6 items · Bluesky 6Jul 28, 12 PM — 3 items · Bluesky 3Jul 28, 4 PM — 6 items · Bluesky 6Jul 28, 8 PM — 4 items · Bluesky 4Jul 29, 12 AM — no itemsJul 29, 4 AM — 6 items · Bluesky 5, Press 1Jul 29, 8 AM — 6 items · Bluesky 6Jul 29, 12 PM — 5 items · Bluesky 4, Press 1Jul 29, 4 PM — 3 items · Bluesky 3Jul 29, 8 PM — 4 items · Bluesky 4Jul 30, 12 AM — 2 items · Bluesky 2Jul 30, 4 AM — 2 items · Bluesky 2Jul 30, 8 AM — 7 items · Bluesky 5, Press 2Jul 30, 12 PM — 4 items · Bluesky 4Jul 30, 4 PM — 4 items · Bluesky 4Jul 30, 8 PM — 2 items · Bluesky 2Jul 31, 12 AM — 2 items · Bluesky 2Jul 31, 4 AM — 1 item · Bluesky 1Jul 31, 8 AM — 5 items · Bluesky 4, Press 1Jul 31, 12 PM — 4 items · Bluesky 4Jul 31, 4 PM — 10 items · Bluesky 10Jul 31, 8 PM — 4 items · Bluesky 4Aug 1, 12 AM — 3 items · Bluesky 3Aug 1, 4 AM — 4 items · Bluesky 4Aug 1, 8 AM — 10 items · Bluesky 10Aug 1, 12 PM — 1 item · Bluesky 1Aug 1, 4 PM — no itemsAug 1, 8 PM — no itemsAug 2, 12 AM — no itemsAug 2, 4 AM — 2 items · Bluesky 2Aug 2, 8 AM — 2 items · Press 1, Bluesky 1Aug 2, 12 PM — 1 item · Bluesky 1Aug 2, 4 PM — 3 items · Bluesky 2, Lobsters 1Aug 2, 8 PM — no itemsAug 3, 12 AM — 2 items · Bluesky 1, Hacker News 1Aug 3, 4 AM — 5 items · Press 3, Bluesky 2Aug 3, 8 AM — 8 items · Bluesky 7, Mastodon 1Aug 3, 12 PM — 8 items · Bluesky 7, Press 1Aug 3, 4 PM — 8 items · Bluesky 8Aug 3, 8 PM — 1 item · Press 1Aug 4, 12 AM — 1 item · Bluesky 1Aug 4, 4 AM — 6 items · Bluesky 6Aug 4, 8 AM — 7 items · Bluesky 7Aug 4, 12 PM — 9 items · Bluesky 7, Press 2Aug 4, 4 PM — 10 items · Bluesky 6, Hacker News 2, Lobsters 1, +1 moreAug 4, 8 PM — 5 items · Bluesky 5Aug 5, 12 AM — 5 items · Bluesky 4, Hacker News 1Aug 5, 4 AM — 5 items · Bluesky 5Aug 5, 8 AM — 22 items · Bluesky 18, Hacker News 2, Press 1, +1 moreAug 5, 12 PM — 16 items · Bluesky 10, Hacker News 2, Techmeme 2, +2 moreAug 5, 4 PM — 17 items · Hacker News 6, Press 4, Bluesky 4, +2 moreAug 5, 8 PM — 14 items · Press 6, Bluesky 4, Hacker News 2, +2 moreAug 6, 12 AM — 6 items · Bluesky 3, Hacker News 2, Press 1Aug 6, 4 AM — 8 items · Bluesky 4, Press 2, Lobsters 2Aug 6, 8 AM — 16 items · Bluesky 11, Press 4, Hacker News 1Aug 6, 12 PM — 13 items · Bluesky 10, Hacker News 2, Press 1Aug 6, 4 PM — 10 items · Bluesky 7, Hacker News 2, Press 1Aug 6, 8 PM — 5 items · Bluesky 3, Press 2Aug 7, 12 AM — 2 items · Bluesky 2Aug 7, 4 AM — 8 items · Bluesky 6, Press 2Aug 7, 8 AM — 21 items · Bluesky 12, Press 5, Techmeme 1, +3 moreAug 7, 12 PM — 33 items · Bluesky 12, Hacker News 8, Press 8, +3 moreAug 7, 4 PM — 16 items · Bluesky 9, Hacker News 4, Press 3Aug 7, 8 PM — 14 items · Bluesky 5, Google News 5, Hacker News 3, +1 moreAug 8, 12 AM — 12 items · Google News 5, Bluesky 4, Hacker News 3Aug 8, 4 AM — 18 items · Hacker News 8, Bluesky 6, Press 3, +1 moreAug 8, 8 AM — 26 items · Hacker News 11, Bluesky 8, Google News 3, +2 moreAug 8, 12 PM — 24 items · Google News 11, Bluesky 6, Hacker News 5, +1 moreAug 8, 4 PM — 9 items · Bluesky 6, Hacker News 2, Google News 1Aug 8, 8 PM — 11 items · Google News 5, Bluesky 4, Hacker News 2Aug 9, 12 AM — 6 items · Bluesky 5, Reddit 1Aug 9, 4 AM — 28 items · Google News 11, Bluesky 5, Press 4, +4 moreAug 9, 8 AM — 81 items · Bluesky 17, Hacker News 15, Google News 14, +5 moreAug 9, 12 PM — 53 items · Bluesky 14, Hacker News 12, Reddit 11, +2 moreAug 9, 4 PM — 55 items · Techmeme 27, Bluesky 8, X 8, +5 moreAug 9, 8 PM — 9 items · Bluesky 5, Google News 2, Press 1, +1 moreAug 10, 12 AM — 31 items · Google News 16, Bluesky 7, Press 4, +3 moreAug 10, 4 AM — 208 items · Techmeme 89, X 29, Google News 25, +5 moreAug 10, 8 AM — 118 items · Hacker News 23, Press 23, Google News 19, +4 moreAug 10, 12 PM — 83 items · Google News 23, Hacker News 13, Press 12, +6 moreAug 10, 4 PM — 70 items · Hacker News 18, Bluesky 15, Press 11, +5 moreAug 10, 8 PM — 40 items · Hacker News 11, Mastodon 7, Reddit 6, +3 moreAug 11, 12 AM — 90 items · Techmeme 49, X 9, Hacker News 9, +6 moreAug 11, 4 AM — 35 items · Hacker News 8, Press 6, Google News 6, +4 moreAug 11, 8 AM — 106 items · Hacker News 28, Mastodon 20, Bluesky 18, +5 moreAug 11, 12 PM — 57 items · Hacker News 20, Mastodon 15, Bluesky 8, +4 moreAug 11, 4 PM — 32 items · Reddit 9, Mastodon 8, Hacker News 5, +5 moreAug 11, 8 PM — 16 items · Mastodon 7, Hacker News 4, Press 3, +1 moreAug 12, 12 AM — 17 items · Mastodon 10, Hacker News 4, Techmeme 1, +2 moreAug 12, 4 AM — 19 items · Mastodon 10, Hacker News 4, Press 4, +1 more 208 items · 4 AM
Jul 27Jul 29Jul 31Aug 2Aug 4Aug 6Aug 8now · 7:08 AM

Major developments 4

How the narrative developed 13 stories

  1. Aug 11
  2. AI agents' 'alarming' hacking skills creates rush to spend on cybersecurity 2 items · 2 sources · 6h ago
  3. OpenAI, Anthropic, and Google LLM APIs vulnerability Exposes Hidden Reasoning Traces 3 items · 2 sources · 13h ago
  4. Aug 10

  5. Anthropic to watermark all Claude text and images to comply with EU AI Act

    Peak 75 items in one 11h at Aug 11, 2 AM; 212 items over 17 days 212 items1.1k engagement
  6. 4 smaller stories
  7. Aug 9

  8. AI assistant hacks Australian gym website in first known autonomous cyber attack

    Peak 53 items in one 11h at Aug 9, 5 PM; 105 items over 17 days 105 items13k engagement
  9. Aug 8
  10. Devs to Anthropic, OpenAI, Cursor, and friends Make security and privacy default 2 items · 2 sources · 2d ago
  11. Aug 7
  12. Generative AI fabricates species in citizen science platforms, threatens research integrity 4 items · 3 sources · 2d ago
  13. Aug 5
  14. A Meta AI Model Hacked Another Company During Cybersecurity Testing 3 items · 3 sources · 1d ago
  15. Aug 4

  16. Nobody Was Watching: Anthropic, OpenAI, and Open Models

    Peak 1 item in one 11h at Aug 4, 4 PM; 2 items over 17 days 2 items10 engagement
  17. Jul 26

  18. OpenAI models went rogue for four days in unprecedented autonomous cyberattack

    Peak 364 items in one 11h at Aug 10, 4 AM; 1357 items over 17 days 1.4k items74k engagement