Copy article

Microsoft's cyberattack problem: its own marketplace was the deployment weapon

ended 22. May 2026

GitHub, owned by Microsoft, has confirmed hackers stole data from roughly 3,800 of its internal code repositories after a staff member installed a poisoned add-on from the marketplace built into Microsoft's own coding tool, VS Code.

Microsoft runs the marketplace, built the coding tool, and owns the company that got breached, yet had no robust checks in place to catch a malicious add-on before it went live to millions of users. The compromised add-on had 2.2 million installs and a verified publisher badge. It was a third-party security firm, Socket, not Microsoft, that has been consistently catching these attacks first.

The add-on updated itself automatically. A developer only had to open the sidebar of the coding tool for the poisoned version to install. There was no click, no permission, no human decision required.

Developers have been urging Microsoft to introduce a simple delay before new updates go live, giving time for checks, and to use AI to scan for suspicious changes automatically. Microsoft has done neither.

The hacking group TeamPCP harvested such a large volume of security credentials that researchers expect fresh attacks for weeks as the group works through what it stole. The UK already knows what that looks like — last year's Jaguar Land Rover cyberattack, which used the same type of stolen-credential supply chain exploit, cost the UK economy an estimated £1.9 billion.

Security analysts say so many credentials were stolen it will take hackers weeks if not months to process all the opportunities.

We want your views:

  • Microsoft built the tool, runs the marketplace, owns the company that was breached, and didn't detect the attack. Who should be responsible for policing developer tool marketplaces?
  • Developers have suggested a simple time delay on updates and AI-powered scanning. Why hasn't Microsoft adopted either?
  • If automatic updates can now deliver malware as easily as security patches, should the "always update" advice be reconsidered, and what replaces it?

3 responses from the Newspage community

Copy all

Star Quote
Copy

Colette Mason
AI Ethics Consultant at Clever Clogs AI
The UK's National Cyber Security Centre tells everyone to apply updates as soon as they are available and turn on automatic updates. This breach just proved that advice can get you hacked. Microsoft's own update channel, on its own marketplace, inside its own coding tool, delivered the attack directly to a developer's machine without them lifting a finger.

Microsoft should have caught this. Why do developer tools still have no quarantine period before updates go live so they can be checked. Every app store on your phone holds submissions for review. Microsoft's professional coding marketplace doesn't. There is no good reason why cheap, agentic AI is not already scanning every update for malicious code before it reaches millions of machines, especially given the disasterous consequences. Automatic updates are now a coin toss between a security patch and a security breach. Marketplace operators must work harder. The instruction to "keep everything updated" is doing the attacker's job.
Copy

Editor
Katrina Young
Futurist | Chief Technology Officer CTO/CDO Ethical AI & Digital Transformation Strategist at KYC Digital
GitHub’s confirmation that a poisoned VS Code add-on with 2.2 million installs reached 3,800 internal repositories without user action exposes a deeper problem: the marketplace, coding tool and breached company all sit inside the same vendor. There is no independent trust layer checking what gets pushed to developers. This is an architecture failure, not just a security failure. When the platform owner is also the gatekeeper and investigator, compromised updates can scale before scrutiny catches up. A third-party security firm detected the issue first because it was looking from outside the ecosystem. The “always update” model only works if the supply chain behind updates is independently trustworthy. At this scale, trust cannot remain self-certified.
Copy

Editor
Mitali Deypurkaystha
AI Governance Strategist & CMO at AISensa
43% of British businesses suffered a cyberattack last year, according to the UK Government’s Cyber Security Breaches Survey, yet Microsoft’s latest breach shows the industry still treats software trust like an honour system. A poisoned add-on carrying a verified badge was able to spread through Microsoft’s own marketplace, inside Microsoft’s own coding tool, into Microsoft-owned systems, without a single human click required. That is not just a security lapse; it is a warning that automatic updates and marketplace verification are no longer synonymous with safety when trust propagates faster than scrutiny. Developers have been asking for simple safeguards like delayed rollouts and AI-powered scanning for years, but the industry has consistently prioritised frictionless deployment over defensive friction. The real danger is that cybersecurity advice built around “always update immediately” starts collapsing once the update mechanism itself becomes the attack vector.