John Yu

What John Yu has said on podcasts

4 statements

1. on BitSecPositiveAug 31, 2026· This Week in Startups

   BitSec's competing AI agents detect and attempt to fix vulnerabilities across codebases.

   “the miners, uh, submit different security-focused AI agents, and we run them through a competition of multiple different codebases with known vulnerabilities in them. And then those AI agents look for those vulnerabilities and try to fix them.”

   Listen at 54:46

   Open the episode · Are AI Agents forming "civilizations" or is this just a psy op? | 2332
2. on Multi-agent cybersecurity systemsPositiveAug 31, 2026· This Week in Startups

   Combining multiple agents and models outperforms a single frontier model for vulnerability detection.

   “the combination of agents with models produces a better performing result than just like, you know, one, one supposedly like superhuman model, like Fable.”

   Listen at 58:08

   Open the episode · Are AI Agents forming "civilizations" or is this just a psy op? | 2332
3. on Qwen modelsPositiveAug 31, 2026· This Week in Startups

   Qwen models provide similar vulnerability-detection capability to Opus 4.8 at 80% lower cost.

   “even QWEN models are able to have the same similar detection capabilities as Opus 4.8 with 80% cheaper because they're open-weight models.”

   Listen at 1:01:50

   Open the episode · Are AI Agents forming "civilizations" or is this just a psy op? | 2332

Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.

• Sep 12, 2026Post-9/11 security delivered gains at a steep democratic costThe NPR Politics PodcastSep 12, 2026
• Aug 31, 2026Tech leaders challenge AI agent hype while regional electric aviation takes offThis Week in StartupsAug 31, 2026Guest