Anthropic released Claude Mythos on April 7, 2026, and the mean time to weaponize a vulnerability collapsed from 2.3 years to 20 hours. Most Salesforce orgs have not updated their security controls to match.
Beech Horn, Technology Engagement Manager and Architect at Banham Group and author at Salesforce Ben, joins Matt Meyers, Salesforce CTA and CoFounder and CEO of EzProtect, to translate what the Mythos-ready framework from CSA, SANS, OWASP, and [un]prompted means for Salesforce security teams right now.
Three principles that anchor this session? The mean time to exploit a vulnerability has collapsed from 2.3 years to 20 hours, and attackers can now transfer a single exploit laterally across entire classes of software within the same day. Offensive AI tools are already shipping with no safety restrictions, which means the technical barrier for launching sophisticated attacks has dropped to a desperate 16-year-old with a MacBook. Salesforce orgs that rely on perimeter security, outdated CSP whitelists, or NIST NVD as their sole vulnerability source are operating on assumptions that stopped being true this year.
This session is for Salesforce architects, consultants, and security leaders who need to secure their orgs against AI-accelerated exploits, agentic prompt injection, and supply chain compromise before the next audit cycle. Watch this video as your guide to the attack surfaces you have not audited yet.
