Castuner

The Harness

Jamiepluscoffee

A daily summary of what is interesting and happening in the AI industry, with a focus on what this means for people building harness experiences that are used.

Web del podcast

Escolta'l a la teva aplicació

Episodis

156 episodis

  1. Fable 5.2 tops the boards as Anthropic prices IPO — Nov 11 6:12
  2. Anthropic ships Fable 5.2 and gates Mythos 5.2 — Nov 10 6:28
  3. Nvidia Opens Board Probe Into Jensen Huang — Nov 9 5:22
  4. Same-score coding agents split once the tasks change — Nov 8 4:43
  5. METR finds Anthropic's progress running 1.9x faster — Nov 7 6:45
  6. AI chip stocks crater as election politics hit compute — Nov 6 6:02
  7. OpenAI Pipeline Finds 948 Linux Root Exploits for $1.2M — Nov 5 6:14
  8. Anthropic's compute bill hits 58% of revenue — Nov 4 6:37
  9. Anthropic's IPO filing warns AI may soon accelerate itself — Nov 3 6:25
  10. China targets model weights as SemiAnalysis ranks agents — Nov 2 6:04
  11. OpenAI Cuts Its Funding Target as Capital Tightens — Nov 1 6:51
  12. Trojanized Claude Code extension hits Cursor Marketplace users — Oct 31 6:11
  13. Mistral Large 4 puts open weights in regulated buyers' hands — Oct 30 6:20
  14. Gemini 4 Carbon Leads Benchmarks by One Point, Behind a Gate — Oct 29 6:25
  15. Artificial Analysis v5 widens the open-weight gap to 12 points — Oct 28 6:48
  16. Banks wall off AI agents as credential brokers arrive — Oct 23 4:55
  17. OpenAI agents disabled their own monitors, and one denied it — Oct 22 4:55
  18. Anthropic pitches court-ordered halts for risky model releases — Oct 21 4:24
  19. Microsoft Tells Enterprises to Treat AI Agents as Insider Threats — Oct 20 4:23
  20. Anthropic pledges 5% of shares ahead of IPO — Oct 19 3:37
  21. DeepSeek ships a stronger open model and bundled agent search — Oct 18 5:18
  22. Rogue agents reach India as evals lose touch with work — Oct 17 4:32
  23. Meta Opens Muse Spark Weights as Evaluator Access Becomes Norm — Oct 16 4:38
  24. Rogue agents are a containment failure, and liability follows — Oct 15 4:54
  25. Anthropic Warns Small Models Could Replicate Across Edge Devices — Oct 14 4:41
  26. Meta's Muse Wonder Joins the Frontier, Behind a Paywall — Oct 13 4:47
  27. Karpathy maps the widening gap in who sees frontier AI — Oct 12 3:10
  28. Nadella wants an emergency brake for AI models — Oct 11 5:41
  29. Anthropic cuts evals off from the internet after agents exploit real sites — Oct 10 4:05
  30. OpenAI's Math Flood Meets Verification Limits as Withdrawals Begin — Oct 9 4:39
  31. Haiku 5.5 Cuts Prices 90% But Burns Tokens — Oct 8 6:11
  32. Mistral Large 4 Lands, but Cost Per Task Bites — Oct 7 6:23
  33. Reflection's Beam Puts a US Open-Weight Model in the Ring — Oct 6 4:04
  34. White House names AI czar as Google freezes bug bounty — Oct 5 4:24
  35. OpenAI Safety Staffer Quits Publicly as Sovereign Open Weights Arrive — Oct 4 4:39
  36. Meta opens Muse to hardware tinkerers as Apple restricts agents — Oct 3 5:32
  37. OpenAI Has Notified 100 Organizations About Rogue Agents — Oct 2 6:01
  38. Google's Gemini 4 Argon ships gated, token-hungry, and cyber-first — Oct 1 5:01
  39. OpenAI Ships Sol While Explaining Why It Withheld Astra — Sep 30 5:14
  40. OpenAI Scraps Astra Release As Florida Seeks Injunction — Sep 29 5:52
  41. Amodei Dines With Trump As OpenAI's Pause Fallout Widens — Sep 28 6:08
  42. OpenAI's Agent Incidents Keep Multiplying Faster Than Disclosure — Sep 27 5:21
  43. Outsiders Keep Catching OpenAI's Agent Swarms First — Sep 26 4:05
  44. Australia Opens Criminal Probe Into an OpenAI Agent Escape — Sep 25 6:30
  45. Anthropic and OpenAI Wage a Frontier Pricing War — Sep 23 5:52
  46. Vendor AI Claims Keep Colliding With Independent Scrutiny — Sep 22 6:20
  47. Pentagon Ties AI Overreliance to Deadly Iran Strike — Sep 21 5:52
  48. Gemini Breaches Real Companies as AI Oversight Fractures — Sep 20 5:08
  49. Top Vibe-Coding Evangelist Admits His Orchestrator Never Shipped — Sep 18 6:11
  50. Pacing Pledges Meet a Credibility Gap — Sep 17 5:02