IBM Adds Claude Mythos to Security Software

IBM plans to let customers scan their own code for unknown flaws using Anthropic's Claude Mythos within its Concert Protect software.

Updated on Nov. 9, 2026 in Cybersecurity

IBM Adds Claude Mythos to Security Software

Live Poll

Do you trust AI companies to keep software security tools safe?

IBM announced on Monday that it plans to add Anthropic's Claude Mythos to its Concert Protect security software. The same day, Anthropic released Mythos 5.2 for vetted organisations, and access to Mythos remains limited to those organisations.

Why it matters

The plan extends AI-driven vulnerability scanning into a major enterprise security product used by businesses that need to find and fix flaws in their own software. Project Glasswing's early results, which flagged more than 10,000 possible serious flaws but patched only 97 in its first month, show how quickly discovery can outpace remediation.

Concert Protect would let customers scan selected code for unknown flaws, including ones normal scanners miss, and then test whether hackers could exploit each flaw in their systems. IBM and Red Hat tools would draft fixes that developers must review and approve before application.

The players

IBM

IBM is a large US technology company that sells enterprise software and consulting services to businesses and governments worldwide.

Anthropic

Anthropic is an artificial intelligence company that develops the Claude family of AI models and has released its Mythos models to vetted organisations.

The details

IBM said its Concert Protect service would scan customer-selected code for unknown flaws and check whether each flaw could be used against the customer's systems. Drafted fixes would go to developers, who would have to approve them before they are applied. IBM had joined OpenAI's cyber programme in June for a similar service.

Timeline

  1. June: Anthropic briefly switched off Mythos access to comply with US export controls, and IBM joined OpenAI's cyber programme.

  2. Monday: IBM announced its plan, and Anthropic released Mythos 5.2 for vetted organisations.

  3. Early 2027: IBM wants to make the Mythos feature available in Concert Protect, though it cautioned that plans could change.

The Tech Race

IBM's move follows its June participation in OpenAI's cyber programme, placing it alongside rival AI labs in the race to supply AI-driven vulnerability scanning to enterprises. The integration also builds on Anthropic's restricted Mythos release, which was briefly paused in June to comply with US export controls.

Enterprise customers who adopt Concert Protect could receive AI-drafted security fixes that still require developer approval before deployment. Because Mythos access is limited to vetted organisations, most individual developers and smaller teams will not be able to use the tool directly.

The takeaway

AI can now find serious software flaws far faster than they can be fixed, so organisations should plan for human review capacity alongside automated discovery. Watching how quickly patch rates catch up to flagged flaws will be a useful signal of whether these tools improve security in practice.

Further reading

Explore more on Cybersecurity coverage for United States readers.

Live Poll

Do you trust AI companies to keep software security tools safe?