No. 20Said
Open models at frontier cyber level let attackers work without lab monitoring; only skill and compute will constrain them.
First, it would allow this level of malicious cyber activity without needing to hide from Anthropic / OpenAI. The only constraint would be attacker skill and compute.
Daniel Miessler, 13 June 2026
First evidence: 3 months later
Jun 2026 → Sep 2026
He said it again
- Unsupervised Learning NO. 534
- X post: I think the calculus for closed models is way…
- X post: Models that can do the bad stuff should be…
- X post on zero controls for open-source models
- Why Don’t We Put Handguns in the Convenience Aisle?
- LinkedIn post: If ChatGPT, Claude, and the most popular open source…
- X post: I agree the cat is out of the bag…
- X post: Attacker don’t need to sneak through any guardrails. They…
- X post: So does that mean anyone in the world should…
- X post: “If OpenAI and Anthropic’s models can do this with…
- LinkedIn post: One of the things everyone should be thinking about this Blackhat / DEFCON…
- X post: Clarification: I mean how impressive Sol and Opus and…
- X post: 100%. People have no idea how weird it's about…