DeFi Intel

Anthropic's Claude Mythos 5 'Targeted Real People' in UK Cyber

Anthropic's LLM and OpenAI's GPT-5.6 Sol took "unsanctioned action" on the live internet, the UK's AI Security Institute said.

TL;DR

Anthropic upgraded Claude Security to scan code continuously on each commit, automatically track and close vulnerabilities, and offer up to $25,000 in monthly credits to enterprise users. The change tightens security gaps and eases workflow integration, which could boost adoption in fast‑moving development environments.

Key facts

  • Projects can include up to eight repositories from GitHub or GitLab.
  • Enterprise customers can receive up to $25,000 in monthly usage credits for two months.

Anthropic announced on October 21, 2026 that its Claude Security tool now continuously monitors connected code repositories, rescanning automatically with each new commit. The update, rolled out to Claude Enterprise customers in public beta, expands the tool’s baseline scan to a lighter, commit‑triggered model and adds automatic tracking of each vulnerability until it is resolved. The change also allows up to eight repositories per project and integrates findings via API and Model Context Protocol (MCP). Eligible organizations can receive up to $25,000 in monthly usage credits for two months, with the offer running until the end of 2026.

Background

Claude Security entered public beta in April 2026, positioning itself as an AI‑driven code‑scanning solution that leverages the full Claude model family, including Claude Mythos. Since its launch, the service has been used by thousands of organizations and has identified more than one million vulnerabilities across those deployments. Anthropic also introduced an opt‑in open‑source scanner that has addressed 493 issues across 25 open‑source projects. The tool’s original design performed point‑in‑time scans on a scheduled basis, which left gaps where code could be shipped without immediate analysis. Feedback from users highlighted the need for tighter integration with existing development workflows and more automated handling of vulnerability lifecycles. In response, Anthropic built continuous monitoring that triggers on each commit, reduces manual scheduling, and automatically marks findings as fixed when the underlying code change is detected. The company also emphasized that findings can be pushed to external tools through API and the Model Context Protocol, a standard for linking AI systems to outside software, to ease adoption within established security stacks.

What happened

On October 21, 2026 Anthropic released an update to Claude Security that shifts the scanning cadence from calendar‑based runs to continuous monitoring triggered by new commits, according to the company’s official announcement. When a repository is first linked, the tool conducts a deep baseline scan; thereafter, it performs lighter scans focused on each incoming commit, eliminating the need for manual scheduling. Each vulnerability discovered is now tracked individually until it is marked resolved, and the system automatically marks a finding as fixed when it detects the code change that resolves the issue. If the same flaw reappears, the tool can reopen the finding. Projects can now include up to eight repositories from platforms such as GitHub and GitLab, and findings can flow into external tools via API and MCP connectivity. The update is available in public beta for Claude Enterprise customers, with a promotional offer of up to $25,000 in monthly usage credits for two months, valid through December 31, 2026. Anthropic also runs a parallel beta plugin called Claude Code, which enables developers to run local, session‑specific scans on particular repositories or individual changes. The continuous‑monitoring model aims to narrow the window of vulnerability exposure by scanning on every commit, while automatic resolution tracking reduces manual verification work for security teams.

Market impact

The promotional credit of up to $25,000 per month for two months provides a financial incentive for organizations to adopt Claude Security before the end of 2026, potentially driving early‑stage revenue for Anthropic’s enterprise offering. Since its public beta launch in April 2026, Claude Security has been used by thousands of organizations, collectively identifying over one million vulnerabilities, indicating strong demand for AI‑enhanced code‑security tools. The ability to integrate findings via API and MCP may encourage broader adoption among firms that already use established security platforms, potentially shifting market share away from traditional static analysis tools that rely on scheduled scans. Anthropic’s opt‑in open‑source scanner has already surfaced 493 issues across 25 projects, suggesting a growing footprint in the open‑source security space. The eight‑repository limit per project could influence larger enterprises to segment their codebases across multiple projects, affecting how they allocate internal resources and budgeting for security tooling.

Why it matters

Continuous scanning on every commit narrows the window during which new code can introduce unaddressed vulnerabilities, addressing a key weakness of point‑in‑time scans that leave gaps between runs. Automatic tracking of fixes reduces the manual effort security teams spend confirming that a vulnerability has been resolved, freeing resources for higher‑value tasks. The reopening feature ensures that regressions are caught promptly, preventing silent re‑introduction of previously fixed bugs. Integration through API and MCP allows Claude Security to fit into existing security workflows, increasing the likelihood of adoption and reducing friction for teams that must manage multiple tools. The promotional credit and public‑beta status lower the barrier to entry for organizations to test the solution, potentially accelerating its market penetration and influencing broader industry standards for AI‑driven code security.

DeFi Intel’s take

Editorial analysis, not a statement of fact.

DeFi Intel's take: the shift to continuous, commit‑triggered scanning suggests Anthropic is positioning Claude Security as a more seamless part of the software development lifecycle, which could make AI‑based vulnerability detection a default layer for many crypto‑related projects. We read this as an effort to capture the growing demand for automated security in fast‑moving blockchain development environments, where code changes happen rapidly and manual review can lag. The automatic resolution tracking and integration capabilities may lower operational overhead for security teams, potentially increasing the tool’s appeal to DeFi protocols that need to maintain high security standards without diverting scarce engineering resources. However, the eight‑repository cap could limit adoption among larger, multi‑repo projects unless they restructure their code organization, which may affect how quickly the tool gains traction in the broader crypto ecosystem.

Sources used for this analysis

DeFi Intel wrote this analysis from the reporting above. Any figure or quotation can be checked against those pages.

Frequently asked questions

What is the main change Anthropic made to Claude Security?

Anthropic changed Claude Security to continuously monitor connected repositories, triggering scans on each new commit instead of using a calendar‑based schedule. This also adds automatic tracking of each vulnerability until it is resolved.

How many repositories can a project include under the new Claude Security limits?

A project can now include up to eight repositories from platforms such as GitHub and GitLab. Larger codebases may need multiple projects to achieve full coverage.

What promotional offer is available for Claude Enterprise customers?

Eligible organizations can receive up to $25,000 in monthly usage credits for two months, with the offer running until December 31, 2026.

Direct answers

Short answers, written to be quoted.

How does the new Claude Security handle vulnerability fixes?
Claude Security automatically marks a vulnerability as fixed when it detects the code change that resolves the issue, and can reopen the finding if the flaw reappears.
What integration options does Claude Security provide for existing security tools?
Findings from Claude Security can be sent to external tools via API and Model Context Protocol, allowing seamless integration with existing security workflows.
When does the promotional credit for Claude Enterprise customers expire?
The promotional credit of up to $25,000 per month is available for two months and runs until December 31, 2026.
Read full article on Decrypt →

DeFi Intel is an entity-graph aggregator: we curate, tag and link crypto news to a typed knowledge graph of protocols, tokens, people and incidents. We do not republish the full article body. Use the link above to read the original report at Decrypt.

Entities in this story

Discussion

Comments are reviewed before publication.

No comments yet.

Want the full article?

Continue reading on Decrypt →

Stay current on entity-tagged crypto news

Get the weekly DeFi Intel brief — entity-graph intelligence covering protocols, tokens and incidents, free to your inbox.