America already lost one AI race | TWiAI Ep 23

This Week In Startups is made possible by:

PAYPAL OPEN

Today’s show:

This week, our panel of AI experts — Anand Kannappan (Patronus AI), Ori Goshen (AI21 Labs), and Alex Finn (Henry Intelligent Machines) — join special guest host Lon Harris to debate whether Washington will move to formally limit powerful foreign AI models like Moonshot’s Kimi K3 and Z.ai’s GLM-5.2.

What does “winning the AI race” actually mean on a practical level? Why is cybersecurity, not benchmarks, likely the actual battleground. And why did Hugging Face have to use GLM-5.2 to investigate their recent breach rather than a frontier US model like Fable 5?

PLUS why agents still fail so much on long horizon tasks, why proper model routing has such a HUGE impact on enterprise AI bills, and we take a look at Alex’s hacked digital typewriter he uses to code without distractions.

Guests:

Anand Kannappan on X: https://x.com/anandnk24

Patronus AI: https://www.patronus.ai/

Ori Goshen on X: https://x.com/origoshen

AI21: https://www.ai21.com/

Alex Finn on X: https://x.com/AlexFinn

Alex’s Vibe Coding Academy: https://www.skool.com/vibe-coding-academy/about

Relevant Links:

Axios: “The Secret Trump Administration Battle to Fight Chinese AI”: https://www.axios.com/2026/07/20/ai-us-china-open-source-kimi

Chamath post on US AI options: https://x.com/chamath/status/2079457219892871458?s=20

Brian Armstrong post on spend vs. token usage: https://x.com/brian_armstrong/status/2070670644577280109

Hugging Face: Security incident disclosure: https://huggingface.co/blog/security-incident-july-2026

VentureBeat Hugging Face coverage: https://venturebeat.com/security/safety-guardrails-blocked-hugging-faces-defenders-not-the-attacker-when-an-ai-agent-breached-its-systems

David Sacks post on cyber guardrails: https://x.com/DavidSacks/status/2078984980588531855

Databricks Unity AI Gateway: https://www.databricks.com/product/artificial-intelligence/unity-ai-gateway

Ramp’s Router: https://ramp.com/router/

Timestamps:

0:00 Will the US limit foreign open source models?

5:40 Are the American frontier AI labs in danger of bankruptcy?

7:15 The risks of over-regulating AI

10:20 What does it mean to win the AI race?

17:08 Is self-hosting agentic Chinese models safe?

20:08 Why agents still fail when models are so smart

25:28 Simulating the digital world for agents

36:51 Why picking the right model for the task is so hard

46:29 Alex is developing "ambient intelligence"

52:35 What we know about the Hugging Face breach

53:46 Are safety guardrails kneecapping our best models?

Subscribe to the TWiST500 newsletter: https://ticker.thisweekinstartups.com

Check out the TWIST500: https://www.twist500.com

Subscribe to This Week in Startups on Apple: https://rb.gy/v19fcp

Follow Lon:

X: https://x.com/lons

Follow Alex:

X: https://x.com/alex

LinkedIn: ⁠https://www.linkedin.com/in/alexwilhelm

Follow Jason:

X: https://twitter.com/Jason

LinkedIn: https://www.linkedin.com/in/jasoncalacanis

Check out all our partner offers: https://partners.launch.co/

Great TWIST interviews: Will Guidara, Eoghan McCabe, Steve Huffman, Brian Chesky, Bob Moesta, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland

Check out Jason’s suite of newsletters: https://substack.com/@calacanis

Follow TWiST:

Twitter: https://twitter.com/TWiStartups

YouTube: https://www.youtube.com/thisweekin

Instagram: https://www.instagram.com/thisweekinstartups

TikTok: https://www.tiktok.com/@thisweekinstartups

Substack: https://twistartups.substack.com