CommBank Releases Free AI Tool to Find Software Flaws

The CodeCheck tool guides AI coding agents to examine software and flag potential weaknesses for engineers to review.

Updated on Nov. 18, 2026 in Cybersecurity

CommBank Releases Free AI Tool to Find Software Flaws

Live Poll

Do you trust AI tools to help keep the software you use safe?

Commonwealth Bank has released CodeCheck, a free AI security tool for identifying software vulnerabilities, designed for businesses using OpenAI Codex and Anthropic Claude Code. The release comes as the Australian Cyber Security Centre notified organisations of potential malicious activity more than 4,400 times in 2025-26.

Why it matters

CommBank built the capability to help strengthen its own defences and wants to share what it has learned with others. The Australian Signals Directorate expects AI-driven tools to shorten the window in which vulnerabilities are exploited further.

CommBank's systems are powered by around 2.5 billion lines of code, and some vulnerabilities were exploited within 48 hours of patches or mitigation advice being released. The tool's effectiveness against those timelines has not been independently reported.

The players

Commonwealth Bank

Commonwealth Bank of Australia is one of the country's largest banks, headquartered in Sydney and known for its major digital banking operations.

Australian Cyber Security Centre

The Australian Cyber Security Centre is the federal government agency that leads national cyber threat monitoring and advises organisations on defending against cyber attacks.

Australian Signals Directorate

The Australian Signals Directorate is the federal intelligence agency responsible for signals intelligence and cyber security, including protecting Australian networks.

The details

CodeCheck guides AI coding agents through a consistent process for examining software and presenting potential weaknesses for human review, and businesses can supply information about how their software is meant to function to give the AI more context. Engineers remain responsible for reviewing findings and deciding what action to take. The tool is available as a free, one-time release through GitHub and requires a supported AI coding agent.

Timeline

  1. 2025-26: The Australian Cyber Security Centre issued more than 4,400 proactive notifications of potential malicious activity.

  2. 17 November 2026: Commonwealth Bank released CodeCheck.

The Tech Race

CodeCheck extends the use of OpenAI Codex and Anthropic Claude Code from writing software into auditing it for security weaknesses. It positions a major Australian bank within the wider push to apply AI agents to cybersecurity defence.

Businesses using CodeCheck need a supported AI coding agent and may incur separate AI model and service costs. Engineers will still need to review each finding before deciding whether any change is required.

The takeaway

AI-assisted code review can surface weaknesses earlier in development, but human engineers remain responsible for judging which findings matter. With exploitation sometimes occurring within 48 hours of a patch, earlier detection is becoming more valuable.

Further reading

For more on how AI is changing digital defence, see Cybersecurity coverage.

Live Poll

Do you trust AI tools to help keep the software you use safe?