Summary:
Proofpoint's Threat Research team has identified a new example of the AI weaponization trend where cybercriminals are exploiting an AI-powered website creation app called Lovable to generate and host malicious sites for credential phishing, malware distribution, and fraud. The platform, designed for users with limited web design knowledge, has been repurposed by threat actors to quickly create and deploy sophisticated phishing websites for their campaigns. The report highlights several examples, including Tycoon PhaaS campaigns that mimic Microsoft authentication pages to steal credentials and MFA tokens using an Adversary-in-the-Middle (AiTM) technique.

The fraudulent websites used in this campaign create an avenue for delivering malware loaders and their subsequent post-compromise tools, as well as cryptocurrency wallet clippers. Proofpoint identified multiple instances where a DLL trojanized with DOILoader dropped zgRAT. Additionally, criminals are using Lovable to create fake websites impersonating shipping companies like UPS to harvest credit card and personal information for more financially motivated endeavors. Proofpoint identified that a campaign impersonating UPS used a malicious site that was based on a public, "remixable" template named "ups-flow-harvester," which allowed other threat actors to easily repurpose it for their own campaigns to collect personal and financial information. These campaigns often use redirectors and CAPTCHAs to bypass security measures and funnel users to the fraudulent pages. While the blog focuses on activity observed in email, Proofpoint researchers have also observed Lovable URLs abused in SMS data, including investment scams and banking credential phishing. The abuse of Lovable was previously highlighted in April 2025 by Guardio in an April 9th blog post.

Security Officer Comments:
The exploitation of legitimate, publicly available tools like Lovable demonstrates a concerning increase in the abuse of AI in cybercrime. As access to advanced technology becomes more commercialized and PhaaS kits become even more prevalent, even script kiddies can launch highly convincing and scalable attacks. These fraudulent websites create a convenient avenue for delivering malware loaders and post-compromise tools, as well as cryptocurrency wallet clippers. The use of a platform with no apparent guardrails for detecting malicious content allows threat actors to rapidly create and modify their infrastructure, making it difficult for traditional security measures to keep up. This trend underscores the importance of a human-centric security approach that focuses on training users to recognize social engineering tactics as they become more authentic with new technology. Guardio notes that this platform is simple to employ and impressively effective.

“Proofpoint reported its findings to Lovable, who then matched them with a cluster of credential phishing previously discovered by Lovable's Trust and Safety team, as well as novel malicious sites. One credential phishing cluster with hundreds of domains was taken down by Lovable the same week. The company said it has also recently implemented AI-driven security protections to attempt to prevent threat actors from developing websites that enable fraudulent activity. According to the company, in July 2025, Lovable introduced both real-time detections to prevent creation of malicious websites as users prompt the tool, and automated daily scanning of published projects to flag potentially fraudulent projects. Lovable said it also plans to release additional security protections pertaining to user accounts to identify fraudulent activity and block malicious users proactively already this fall.” (Proofpoint 2025)

Suggested Corrections:
IOCs are available here.

General Phishing Recommendations:
Users should always be cautious of individuals or organizations that ask for personal information. Most companies will not ask for sensitive data from their customers. If in doubt, users should verify with the company itself to avoid any potential issues.

Users should always take a close look at the sender’s display name when checking the legitimacy of an email. Most companies use a single domain for their URLs and emails, so a message that originates from a different domain is a red flag.

As a general rule, users should not click links or download files even if they come from seemingly “trustworthy” sources.

Check for mismatched URLs. While an embedded URL might seem perfectly valid, hovering above it might show a different web address. In fact, users should avoid clicking links in emails unless they are certain that it is a legitimate link.

Users should always be on the lookout for any grammatical errors and spelling mistakes. Legitimate companies will often employ proofreaders and editors who ensure that the materials they send out are error-free.

Users should not be frightened or intimidated by messages that have an alarmist tone. They should double check with the company if they are uncertain about the status of their accounts.

Phishing emails are designed to be sent to a large number of people, so they need to be as impersonal as possible. Users should check whether the message contains a generic subject and greeting, as this can be a sign of a phishing attempt.

Although not every end user has access to advanced anti-phishing software, they can still use the built-in protection of their email clients to filter messages. One example is setting the email client to block all images unless approved.

Legitimate companies will never send confirmation emails unless there are specific reasons for doing so. In fact, most companies will avoid sending unsolicited messages unless it’s for company updates, newsletters, or advertising purposes.

Users should always take the context of an email or message into account. For example, most online accounts do away with viewable member numbers, so users should be wary if they receive emails containing a “member number” for services that generally don’t use them.

It is important to take note of unusual information in the text of the message. Any mentions of operating systems and software that are not typically used by consumers can often be indicators of a phishing attempt.

If it seems suspicious, it probably is. Users should always err on the side of caution when it comes to sending out personally identifiable information through messages and emails.

Link(s):
https://www.proofpoint.com/us/blog/threat-insight/cybercriminals-abuse-ai-website-creation-app-phishing