Trust and safety UX: Scareware Blocker

How we warned people about scareware in Edge - strongly enough to protect them, without amplifying fear.

My role

UX design for scareware and security intervention flows, with PMs, engineers and security stakeholders.

Focus

End-to-end ownership, security

Outcome

Shipped in Edge, announced at Ignite and covered in the press since launch.

Context

Scareware is designed to exploit fear, urgency and confusion to coerce users into unsafe actions. In a browser, poor interventions can fail to protect users or damage trust.

This case study brings together multiple trust and safety surfaces I worked on across Edge, including Scareware Blocker, Browser Essentials security flows, threat messaging, and related error and recovery states.

My role and responsibility

I was responsible for the UX design of key scareware and security intervention flows, with a focus on messaging, severity calibration and recovery paths. I worked closely with PMs, engineers and security stakeholders to balance user protection, false positives and trust. This included:

  • Threat intervention flows
  • Messaging and language tone
  • Severity and visual emphasis
  • Error and recovery states
  • Cross-team reviews with security partners

The problem - fear vs trust

The central challenge was deciding when to intervene, how strongly, and how to help users recover safely - without amplifying fear.

Under-warning

Users fall victim to scams

Over-warning

Users stop trusting the browser

Ambiguous warnings

Panic and unsafe decisions

Severity calibration

We already had some empty state and error page warnings, but they didn’t feel aligned. I was conscious of respecting those pages to retain trust - and it was also a chance to address them.

Not all threats deserve the same visual weight. I needed to convey a potential risk rather than a hard block, but something more important than an informational warning like a 404 error.

Messaging that reassures

Finding messaging that reduced panic and avoided alarmist language was important. The AI could sometimes raise false positives, so I left the final choice to the user on whether to proceed.

User testing showed it helped most to keep a screenshot of the destination site on screen, so people could make a more informed decision. Language was intentionally designed to slow users down rather than escalate urgency.

Design decisions were shaped by security accuracy limits and legal review, which constrained both language and interaction patterns.

Outcomes and learnings

Scareware Blocker has been covered in several press releases since launch, and it’s great to read about how it’s helping users. This feature matters a lot to me, because it helps vulnerable people stay away from scams.

Check out the Microsoft Blog article on Scareware