normative

Open source maximizes eyeballs on dangerous AI

If a technology is genuinely dangerous, you would want more eyeballs on it, not fewer; open source maximizes transparency, lets people identify misuse and jailbreaks faster, and enables academics with access to weights to produce breakthroughs (and better guardrails) that benefit all frontier models — which is why Zuckerberg's open-sourcing of capable models like Llama-3-70B is valuable.

normativepending

Speaker

Aravind Srinivas

Evidence Quote

“You can identify more ways the systems can be misused faster and build the right guardrails against it too.”

Source

Aravind Srinivas— Lex Fridman Podcast
Created: 6/13/2026, 3:26:05 AM

Your notes

Loading notes...