Anthropic's announcement expanding its Cyber Verification Program. The companion cites it for one figure: by July 2026 defenders using Claude Mythos models had found at least 129,000 verified vulnerabilities, more than 33,000 of them critical or high-severity. These directed defender agents are among the enemies of parasites in the datasphere's food web.
- Cited in
- the companion paper
- Used for
- Strand
- Cybersecurity & AI-enabled attack
Cite
@misc{anthropic2026cvp,
title = {Expanding the Cyber Verification Program},
author = {Anthropic},
year = {2026},
howpublished = {Anthropic News},
url = {https://www.anthropic.com/news/cyber-verification-program},
}