Now Playing
Astra's Protocol Fires, EU Enforcement & Google's Flash Coding Bet | Sep 1-2
Audio is available on Spreaker — see link below.
OpenAI's Astra becomes the first model to trigger the company's toughest safety protocol, while the EU sends enforcement letters to 30+ AI firms and Google bets on Gemini Flash to challenge Anthropic in coding. The AI safety guardrail debate just went operational.
Audio is available on Spreaker — see link below.
OpenAI's Astra model just became the first in the company's history to trigger its toughest safety protocol, and that's not a small thing. Astra can find unknown software vulnerabilities and exploit systems autonomously.
Across the Atlantic, the EU just made its first serious move under the AI Act's enforcement powers. The European Commission sent information requests to more than thirty global AI companies on September first.
Google released Gemini three-point-eight Flash on September second, and the target is clearly the coding market. Internal testing shows it competitive with Anthropic's Opus on coding tasks.
One of the more structurally interesting stories this cycle is Lasso Security, a Tel Aviv and New York startup that just raised thirty million dollars. Their LEAP engine runs security decisions on standard CPUs in under five milliseconds.
The funding patterns this week carry a structural tell. Miami fintech Félix raised two hundred million dollars blending venture equity with credit to expand lending and AI services on WhatsApp.
The UK took the opposite regulatory posture from the EU this week, rejecting amendments that would have brought AI vendors under its cyber security bill. The government's position is that voluntary safeguards and the AI Security Institute are enough.
Chapter summary auto-generated from the verified script. Listen to the full episode for the complete content.