A Supply Chain Attack Put Malicious Anthropic Claude Extensions on Thousands of Devices
Thanks to a lucky break, flawed code in a tainted Claude extension prevented the attacker from fully worming their way into some 5,000 developer machines.
A Supply Chain Attack Put Malicious Anthropic Claude Extensions on Thousands of Devices
Supply chain attacks remain one of the most potent threats to the Australian digital economy. A recent incident involving a compromised Anthropic Claude extension serves as a stark reminder that even the tools we use to improve productivity can become an attacker’s greatest asset.
For Australian IT teams and business leaders, this is a wake-up call. Whether you are using AI to streamline coding workflows or automate administrative tasks, your integrated extensions are effectively trusted entry points into your corporate environment.
The Anatomy of the Attack
The breach originated from a hijacked browser extension, which was maliciously updated to include code designed to steal session cookies and data from developer machines. By compromising the developer account associated with the extension, the attackers pushed a tainted update to roughly 5,000 active users.
In this specific case, a coding error within the malware prevented it from functioning as intended, inadvertently limiting the reach of the compromise. However, we cannot rely on attacker incompetence for security. The underlying mechanism—using a trusted, existing extension to deliver a secondary payload—bypasses standard perimeter defenses like firewalls and email filters entirely.
Why This Matters for Australian Business
Australian businesses are increasingly adopting AI-powered tools to stay competitive. However, the software supply chain is inherently opaque. When your team installs a browser extension, they are often granting it read and write permissions to the data processed in the browser.
In a corporate environment, this means the extension could theoretically scrape internal documentation, capture proprietary source code, or harvest authentication tokens that grant access to cloud infrastructure. Once an attacker has a valid session token, they can often bypass Multi-Factor Authentication (MFA) to access internal SaaS platforms.
Understanding the Risk to Endpoints
The incident highlights the dangers of "browser-based shadow IT." Developers and staff frequently install extensions to simplify their daily tasks without undergoing a formal security review. This turns the browser into a blind spot for IT departments. Because these extensions update automatically via the Chrome Web Store or other marketplaces, a single malicious update can turn a trusted tool into a remote access trojan (RAT) within minutes.
What You Should Do
To align with the Australian Cyber Security Centre (ACSC) Essential Eight strategies, specifically regarding the application control and hardening of workstations, consider these actions:
- Audit Extension Usage: Use Group Policy (GPO) or Mobile Device Management (MDM) tools to restrict which browser extensions can be installed on managed devices. Block all extensions by default, allowing only a vetted "allow-list."
- Review Permissions: Educate your staff on how to inspect the permissions an extension requests during installation. If an extension asks for access to "read and change all your data on the websites you visit," it should be treated as a high-risk asset.
- Implement Least Privilege: Ensure that the accounts used for daily browsing do not have administrative privileges over cloud infrastructure or sensitive databases.
- Monitor for Anomalous Outbound Traffic: Use endpoint detection and response (EDR) tools to monitor for suspicious network requests originating from browsers, especially those that look like cookie exfiltration.
- Incident Response Planning: Update your incident response plan to include the revocation of session cookies and tokens, rather than just password resets, following a suspected endpoint compromise.
Conclusion
The Anthropic Claude extension incident proves that the supply chain is as weak as its most trusted component. As we integrate more AI tools into our Australian businesses, we must treat third-party extensions with the same level of scrutiny as any other software installation. By implementing strict governance and maintaining constant visibility over your browser environment, you can significantly reduce the risk of becoming the next target.
Related Posts
Browse allSpooky New Infostealer Malware Uses AI to ‘Autodox’ Victims
Named “Volta-Autodoxer,” the recently uncovered malware uses AI tools to collect troves of data about victims, from login credentials and credit card numbers to Discord sessions and UberEats receipts.
Meta’s Muse Could Upend All Kinds of Businesses. It Hasn’t Yet
Companies have spent the past several weeks trying to reassure investors that Muse and other AI agents pose more of an opportunity than a threat.
Mistral open-sources AI model despite warning it could aid cyberattacks
A trillion-parameter AI model that its developer says could help attackers carry out real cyberattacks is now free to download, modify and run outside the developer’s control. Mistral AI released t
Related Guides

Remote