Vercel’s v0.dev AI Tool Weaponized for Phishing Campaigns Targeting Microsoft, Nike Users

Security firm Cofense has published a warning about a significant and ongoing increase in phishing campaigns built using Vercel's v0.dev generative AI platform, which allows attackers to create convincing credential-harvesting pages impersonating brands like Microsoft, Spotify, Adidas, Nike, Ferrari, and Louis Vuitton with no web development expertise required. The research, published by Infosecurity Magazine, highlights a shift in the phishing threat landscape: AI tools designed for legitimate developers are being systematically abused to democratize high-quality phishing kit creation.

v0.dev and the Phishing Attack Chain: Technical Details

v0.dev is a generative AI code tool provided by Vercel — the cloud platform known for hosting frontend JavaScript applications — that produces complete, styled, functional web applications from plain-text prompts. A user can describe what they want ("a login page that looks like Spotify with a username and password field and a submit button") and v0.dev generates the full HTML, CSS, and JavaScript for a production-ready page, hosts it on Vercel's infrastructure under a vercel.app domain, and makes it immediately accessible over HTTPS.

Cofense's analysts observed attackers using exactly this workflow to produce phishing pages. Because v0.dev handles both code generation and hosting, the attacker does not need to:

• Write HTML, CSS, or JavaScript
• Purchase or configure a web hosting account
• Obtain a TLS/SSL certificate for HTTPS
• Register a lookalike domain (though attackers may do this additionally)
• Buy or configure a phishing kit from underground markets

The resulting pages are served from Vercel's global CDN (Content Delivery Network), giving them fast load times and high availability. They carry valid HTTPS certificates issued to Vercel infrastructure, which passes basic "check for the padlock" security education that many users still rely on. The vercel.app domain carries Vercel's trust reputation with email gateways and DNS filtering systems that have not yet categorized individual subdomains as malicious.

Cofense also documented the use of v0.dev's integration features in more sophisticated campaigns. Vercel's platform offers built-in integrations with Telegram (a messaging platform), AWS (Amazon Web Services), Stripe (a payments platform), and xAI (Elon Musk's AI company). Attackers have used the Telegram integration to route harvested credentials directly to attacker-controlled Telegram channels in real time, creating a lightweight C2 (Command and Control — the infrastructure through which attackers receive data and issue instructions) pipeline without building separate infrastructure. Cloudflare's Cloudforce One team documented an evolution of this technique in which Vercel-hosted campaigns used Telegram for victim filtering — sending stolen credentials to operators who then decided in real time whether a victim's account was worth further attention.

Exploitation Status: Campaigns and Brands Targeted

Cofense observed phishing campaigns actively impersonating the following brands using v0.dev-generated pages:

• Microsoft — fake Microsoft 365 and Azure login pages targeting enterprise credential harvesting
• Spotify — fake streaming login pages, likely targeting password reuse (attackers assume Spotify passwords are reused elsewhere)
• Nike and Adidas — fake e-commerce login pages, targeting payment card data and account takeover
• Ferrari and Louis Vuitton — fake luxury brand account pages, targeting high-net-worth individuals

Campaigns range from simple credential capture pages — clone the login UI, forward credentials to Telegram — to more elaborate flows involving fake email notifications, fake order confirmations, and multi-step "verify your account" flows that increase the perceived legitimacy of the page. The Cofense threat prediction report notes that the ease of restarting a taken-down campaign is a key operational advantage: if Vercel removes a malicious page, the attacker regenerates it with a new v0.dev prompt in minutes, obtaining a fresh subdomain and continuing the campaign with minimal disruption.

Who Is Affected

The immediate targets of v0.dev-based phishing campaigns are end users at organizations whose employees use Microsoft 365, Azure, or any of the consumer brands being impersonated. However, the threat surface extends to:

• Enterprise security teams whose email gateways and URL filtering systems may not yet categorize vercel.app subdomains as suspicious, since most Vercel-hosted traffic is legitimate developer content
• Microsoft 365 administrators and identity teams, since Microsoft credential phishing is the most operationally damaging variant in this campaign set — enterprise email access enables business email compromise (BEC — fraud conducted by impersonating a trusted employee or vendor via email), lateral movement, and downstream data exfiltration
• Security awareness training programs: Organizations whose users have been taught that HTTPS equals safe are specifically vulnerable to the Vercel phishing pattern, since all v0.dev pages carry valid HTTPS certificates

Separately, organizations that use Vercel for their own legitimate web hosting should be aware that brand association with a platform being widely used for phishing may affect how their content is treated by email security scanners going forward.

What You Should Do Right Now

• Block or monitor v0.dev and vercel.app subdomains in email security gateways: Most enterprise mail environments support URL reputation filtering. Add Vercel-generated domains (*.vercel.app) to a monitored or sandboxed category rather than an unconditional block — most Vercel traffic is legitimate — but ensure clicked URLs from Vercel-hosted pages are scanned in a sandboxed browser before rendering to end users.
• Update security awareness training: Explicitly inform users that HTTPS and a padlock icon do not indicate a trustworthy page. Modern phishing infrastructure universally delivers HTTPS — the padlock confirms encryption, not legitimacy. Train users to verify the full domain name, not just the protocol.
• Enable phishing-resistant MFA across Microsoft 365 and Azure: FIDO2 hardware security keys and Microsoft Authenticator passkeys are resistant to real-time phishing relay attacks, where an attacker's proxy site captures both credentials and MFA codes simultaneously. Time-based one-time passwords (TOTP — authenticator apps that generate 6-digit codes) are not phishing-resistant and provide limited protection against this attack pattern.
• Report Vercel-hosted malicious pages directly: Cofense recommends reporting malicious vercel.app pages to Cofense for takedown coordination. Vercel accepts abuse reports at its security disclosure program; direct reports accelerate page removal.
• Check email authentication configurations: Ensure your domain's SPF (Sender Policy Framework), DKIM (DomainKeys Identified Mail), and DMARC (Domain-based Message Authentication, Reporting, and Conformance) records are configured with a DMARC policy of p=reject. This prevents attackers from sending spoofed email appearing to come from your domain, which reduces the effectiveness of campaigns targeting your employees or customers.
• Hunt for Vercel-hosted pages in your recent email logs: Query your email gateway logs for links to *.vercel.app or v0.dev domains in email received by your organization over the past 90 days. Any clicks on those links by users warrant investigation.

Background: Understanding the Risk

v0.dev-based phishing is one expression of a broader trend: AI tools that were built to make legitimate development faster are simultaneously making attacks that previously required skill and infrastructure accessible to low-sophistication threat actors.

The traditional phishing kit — a zip file containing cloned HTML, PHP scripts for credential collection, and configuration files — required a threat actor to acquire a hosting account, set up a web server, configure redirect chains, and maintain infrastructure as takedowns occurred. Sophisticated kits from underground markets sold for hundreds of dollars and were maintained by specialists. AI-powered platforms like v0.dev collapse most of that operational overhead: page generation, hosting, HTTPS, and CDN delivery are provided automatically, the barrier to entry becomes near-zero, and the quality ceiling rises.

This pattern is not limited to Vercel. Research in 2025 and 2026 has documented similar abuse of platforms including Netlify, GitHub Pages, Cloudflare Pages, and Glitch — all platforms that offer free HTTPS hosting with trusted domain reputations. The common thread is that legitimate developer infrastructure makes phishing infrastructure indistinguishable from genuine content at the infrastructure level; differentiation must happen at the content or behavior layer.

From a defensive standpoint, the long-term answer is not to block all these platforms — they host significant legitimate content — but to ensure that credential submission forms on these domains are treated with elevated skepticism, and that phishing-resistant authentication methods reduce the value of successfully harvested credentials.

The Cofense disclosure follows a separate April 2026 incident in which Vercel itself suffered a security breach involving ShinyHunters and OAuth token compromise — a distinct issue from this phishing abuse campaign, but one that has placed Vercel under heightened security scrutiny in recent weeks.

Conclusion

Threat actors are systematically using Vercel's v0.dev AI tool to generate and host convincing phishing pages targeting Microsoft, Spotify, Nike, and other major brands at scale, with no development skill or infrastructure investment required. Security teams should update email gateway rules for Vercel-hosted URLs, retrain users on HTTPS trust assumptions, and prioritize deployment of phishing-resistant FIDO2 MFA across Microsoft 365 and Azure environments — the primary enterprise credential target in these campaigns.