Krux
April 13, 2026
Anthropic Opens Cybersecurity Model to Three Cloud Giants
Published: April 13, 2026 at 12:30 AM
Updated: April 13, 2026 at 12:30 AM
100-word summary
Anthropic's Mythos Preview, a model built to hunt software vulnerabilities, is now accessible through Amazon Bedrock, Google Vertex AI, and Microsoft Foundry. Security teams can point it at their codebases to find exploits, test binaries for weaknesses, or run penetration tests without writing a line of code. Access is still gated (invite-only through AWS in Virginia for now), but the three-cloud rollout signals Anthropic thinks defensive AI is ready to secure critical software. The catch? It stays locked down for safety reasons. Translation: the same model that finds your vulnerabilities could find someone else's.
What happened
Anthropic's Mythos Preview, a model built to hunt software vulnerabilities, is now accessible through Amazon Bedrock, Google Vertex AI, and Microsoft Foundry. Security teams can point it at their codebases to find exploits, test binaries for weaknesses, or run penetration tests without writing a line of code. Access is still gated (invite-only through AWS in Virginia for now), but the three-cloud rollout signals Anthropic thinks defensive AI is ready to secure critical software. The catch? It stays locked down for safety reasons.
Why it matters
Translation: the same model that finds your vulnerabilities could find someone else's.