‘Yahoo Boys’ Telegram Channels Adopt AI Tools and New Tactics for Romance and Impersonation Scams

Users in Telegram channels openly associated with networks of West African cybercriminals known as “Yahoo Boys” have discussed using artificial intelligence tools, including generative AI models and face-swapping and voice-cloning software, to make their scams more convincing and effective.

The term “Yahoo Boys” refers to networks of West African cybercriminals, often younger men in countries like Nigeria, best known for targeting web users in Western nations using social engineering tactics like romance scams, wherein scammers develop fake relationships with their victims that they later exploit to extract their victims’ money and sensitive personal information; many schemes also involve impersonation, as seen in the network’s now-infamous “Nigerian Prince” email scams.

As Open Measures previously observed on 4chan, the posts we identified showed that users in chat groups associated with Yahoo Boys networks have applied AI tools in increasingly sophisticated ways. Today, some Telegram group leaders even offer educational “VIP classes” and “AI tools packages” to subscribers looking to enhance their abilities to scam others.

Tech and cybersecurity reports on the Yahoo Boys have often been sardonic, portraying the network as a “low-level” threat whose limited successes can be attributed to a strategy of targeting naive victims. However, recent reports have detailed the Yahoo Boy network’s expansion into more sinister sextortion scams and overlap with Black Axe, a violent Nigerian criminal syndicate recently pursued by state authorities, INTERPOL, and the FBI, suggesting the network may pose a more serious threat today than it had in the past.

their threat level may have adjusted. Through social media sites, dating apps, and fintech services, the network’s reach has expanded and now takes on more ambitious targets; these and other threats may soon be amplified by the Yahoo Boys’ increasing fluency in AI tools.

Our Approach

To separate explicitly Yahoo Boys-affiliated Telegram channels from other likely cybercriminal marketplaces we monitor on the platform, we limited our analysis to posts shared in Telegram channels and groups in our datasets that included the term “yahoo” in their names. Using the analysis tools in our platform, we identified and analyzed all posts that appeared in those channels since the start of the 2026 calendar year and identified a handful of channels where a majority of the relevant posts we found had appeared.

Caption: An Open Measures Timeline chart lists four Yahoo Boys-affiliated channels identified by Open Measures researchers via wildcard search, with context filters for “yahoo” enabled to target only channels that explicitly claimed Yahoo Boys affiliation in their channel names or usernames. Identified are four, one-way broadcast channels responsible for 6,629 query-matching posts contained in Open Measures datasets in 2026: @yahooformatt1, @format3600, @abra666999, and @lordofupdate. While the most recent of @lordofupdate’s context names do not contain “yahoo”, our researchers surmise that its context names may have been changed from names that did. Upon manual review in Discovery, researchers confirmed @lordofupdate did express an explicit affiliation with Yahoo Boys networks and have chosen to include it in this analysis.

The analysis that follows is based on our review of posts in the channels we identified, but does not represent all Yahoo Boys-affiliated activity on Telegram. Rather, our findings present a partial view at the ways cybercriminals and scammers have looked to AI tools to enhance the effectiveness and scale of their illicit operations.

How Yahoo Boys Are Using AI Tools

The sorts of romance and impersonation scams that Yahoo Boys became notorious for were relatively unsophisticated and required their perpetrators to invest significant amounts of time developing relationships with their targets that can later be exploited. Their success depended on a scammer’s ability to appear genuine and respond to tests of their authenticity, like sending requested photos or answering calls from their victims.

Disarming their victims’ skepticism can sometimes prove difficult for Yahoo Boys scammers, like when a target requests that they participate in a live video call to prove their identity or asks for specific forms of confirmation, like documents. The posts we reviewed from the Yahoo Boys-affiliated Telegram channels in our datasets showed that scammers in those networks have looked to AI tools to pass those checks, potentially enabling them to carry out more convincing scams at greater scales than before.

Throughout 2026, we saw administrators of Yahoo Boys channels recommend increasingly sophisticated AI tools and techniques for using them. Some of the earliest posts we identified from the year recommended free and simple-to-use generative AI programs capable of producing synthetic images, videos, and audio files that could make a scammer’s profile appear more legitimate. But as the year went on, some of those same channels began recommending more advanced AI tools, including programs capable of rendering synthetic material in live video calls.

Open Measures identified posts recommending specific AI tools by name and shared instructions for using them in furtherance of impersonation and romance scams. We’ve listed some of the tools that appeared by name most often in posts in the channels we analyzed and explained how users in Yahoo Boys-affiliated channels proposed using them.

Yahoo Boys’ Synthetic Voice-Cloning and Face-Swapping Strategies Grew More Advanced Through 2026

In the posts we reviewed, users in Yahoo Boys-affiliated Telegram channels most often recommended AI programs that allow users to generate synthetic media. The first mentions of such tools we identified directed channel subscribers toward free and relatively simplistic apps. Later posts pointed users toward more complex programs, including those capable of producing “live” videos.

In order of their appearance in the channels we reviewed, the tools recommended in posts included:

Any Voice Lab, an AI tool allowing users to submit an audio snippet and convert it into one of thousands of available voice templates and accents. The app’s free version includes a character limit and allows a limited number of samples to be made before a user must pay to subscribe, a feature which, admins suggested, might make the tool suboptimal for certain scams (Apr. 27, “Dating Formats and Yahoo Updates”).

FakeYou, a browser-based app for converting short audio and video clips into lip-synced deepfake videos, cited as a “sharp site to clone your celeb voice.” The app’s free tier allows users to produce audio and video more generously than Any Voice Lab, making it a more scalable tool enabling some simple, low-quality use cases for celebrity impersonation scammers (with comparatively few celebrity voice and video templates available (May 10, “Dating Formats and Yahoo Updates”).

Revoice, a similarly simple, audio-only “clone voice” app, mentioned in a post that included a screen recorded walkthrough briefly explaining the tool (May 15, “Dating Formats and Yahoo Updates”).

DreamFace, an app allowing users to convert videos of themselves into simple, lip-synced deepfake videos complete with audio, described by admins suggested for making short videos scammers could send to their victims as identity proof in lieu of answering them on a live call. As the app’s capabilities are somewhat more advanced than earlier tools, the post we identified included a detailed tutorial (Aug 21, “Yahoo Format”):

upload a clear photo with the person’s face looking directly at the camera . . . Choose how the photo will speak . . . Tap generate or create . . . the AI animates the face and syncs the lips to the audio . . . preview the result. If you’re happy with it, save or export the video. Free accounts often add a watermark or limit the number of videos you can create

Caption: A composite image of various screenshots of in-use AI tools, shared in Yahoo Boys channels that offered subscribers tutorials and techniques for voice cloning and face swapping to improve romance and impersonation scams (from left: Revoice, FakeYou, AnyVoiceLab, DreamFace).

Later in 2026, our researchers began to observe posts recommending commercial and enterprise-grade AI tools more frequently, for many of the same use cases mentioned above:

ChatGPT was recommended alongside explicitly detailed text prompts for generating “ultra-realistic” photos of Canadian and Nigerian identification cards held in a person’s hand (likely to submit to online identity verification tests). The example prompts included in posts we identified were schematically formatted and labelled, possibly allowing users to edit their fields and generate other false documents (May 1, “Dating Formats and Yahoo Updates”). A separate post recommended the app for compositing images of scammers’ fake profiles into images featuring Elon Musk, a technique designed to bridge romance scam victims into investment scams described at length in the image below (May 25, “Yahoo Format”).

Snapchat AI, the social media platform’s built-in generative AI writing assistant, which admins recommended for generating “sharp descriptions” for fake profiles – useful as a platform-native way to populate fake profiles with fluent, colloquial English text (June 19, “Yahoo Format”).

Higgsfield, a commercial AI image and video generation app, was suggested as a “site to do face swap” capable of supplementing authentic stolen images and videos by generating additional, inauthentic ones more passably than simpler tools (June 19, “Dating Formats and Yahoo Updates”).

ElevenLabs, a hybrid platform for generating text-to-speech AI voice audio and related video, was suggested for the creation of “realistic male or female” voices with detailed instructions for fine-tuning outputs (Aug 20, “Yahoo Format”):

if you want it to sound more expressive, adjust the voice settings (such as emotion or pacing if available)

Caption: A composite image of Telegram identified in “Yahoo Format,” each of which suggest scam-related use cases for commercial and enterprise-grade AI platforms. Included posts mention ChatGPT for assistance in the “Elon Musk update” (top left), Grok for assistance with Meta Business Suite while recruiting targets for celebrity impersonation scams (bottom left), and Snapchat AI for generating “sharp descriptions” to fill out fake social media profiles (right).

InnoCaption, a free app providing call-captioning services for the deaf and hard-of-hearing, was promoted by admins as a tool for smooth, type-to-text “voice cloning” on live calls (the app also has an AI assist feature, capable of expanding a sentence fragment into a full sentence and converting it into voice almost instantaneously). InnoCaption’s official website lists use cases for audiologists, veterans, and those with disabilities, promising it can help users “make calls with confidence.” It is also FCC-certified, requiring US-based phone numbers and proof of US residency from customers in order to use the platform; for a non-US citizen, accessing the app requires providing fraudulent documentation, which AI tools have made much easier to generate (June 21, “Dating Formats and Yahoo Updates”).

Magicam.ai, an AI virtual camera self-described as “the ultimate real-time face swap solution,” was similarly described in a post by Yahoo Boys admins as a “live video call changer, voice clone all in one,” though posts promoting it noted that it required “powerful hardware support” – ideally, a NVIDIA graphics card – to run properly (Aug. 24, “Dating Formats and Yahoo Updates”).

Pyzra AI Camera, a similar AI virtual camera app, was recommended to help users “face swap on WhatsApp video call” in a post that included a two-minute instructional video voiced by an AI narrator, complete with “step by step” instructions, from installation through Google’s App store to uploading reference images, configuring audio and video, and adjusting aspect ratios to render images correctly (Sep. 10, “Dating Formats and Yahoo Updates”).

Caption: An image composite of materials drawn from two Telegram posts, each referring channel subscribers to AI tools for combined face swapping and audio cloning on live calls with victims. A post from September 10 with a brief video walkthrough attached recommends Pyzra AI Camera (top), while another post from August 24 suggests Magicam.ai (bottom). The latter includes a screenshot from the app’s official site, which lists its technical requirements beneath a banner ad offering “75% Off for a Limited Time!”

Admins in Yahoo Boys Channels Selling ‘VIP Classes’ and ‘AI Tools Packages’

Many of the Yahoo Boys-affiliated channels we identified appeared to operate as storefronts, sharing posts that listed menus of tools and services that users might purchase to enhance their fraudulent and scam activities. In these channels, we saw that admins have sought to capitalize on other Yahoo Boys-affiliated users’ interest in AI tools by offering educational material and specialized programs tailored toward scammers.

For example, a Feb. 21 post in the “Dating Formats and Yahoo Updates” Telegram channel advertised a “CELEB VIP CLASS” that claimed to include an exclusive “AI Tools Package” and guidance on advanced techniques that subscribers could use to make their own impersonation scams more convincing.

We observed the first of these posts on June 7, though the operator shared an additional 20 posts – 11 of which were nearly identical – periodically from June through October. Many of these posts included a Telegram username at the bottom, asking those interested to sign up and coordinate payment via DM. Variants of this post template that we observed offered the class at a fluctuating price point that appeared to be enumerated in local currencies accepted by the operator (“50k 480gh” and “30K (Regular price: 50K),” with “k” and “gh” likely indicating Nigerian naira and Ghanaian cedis).

Caption: A composite of two Telegram posts formatted as advertisements, shared in “Dating Formats and Yahoo Updates” by the channel’s anonymous admin (one of which is signed with the username @loven11). On the left, a post from January 1, 2026 advertises a range of likely scam-related services, many of which appear designed to facilitate other scammers’ cashouts by purchasing “valid gifts cards” they received and routing the funds back to them through pre-verified bank accounts and payment processing services (Venmo, CashApp, PayPal). On the right, a similar post from June 7 follows the same format but now exclusively advertises a paid “CELEB VIP CLASS,” claiming to teach subscribers exclusive techniques for celebrity impersonation scams with an “AI Tools Package Included.” (The “VIP CLASS” post we include above is from June 7, though the earliest post we identified in this format appeared on February 21.)

Through native Telegram search, our researchers confirmed both “Yahoo Format” and “Dating Formats and Yahoo Updates” were still active as of this writing.

Our Take

In September 2024, Telegram announced it would take a stronger stance against cybercrime on its platform, a position the company demonstrated by removing millions of channels suspected of illicit activity throughout 2024 and 2025. Telegram has continued reporting tens of thousands of daily takedowns throughout 2026, though our own research has shown cybercriminals remain resilient to the platform’s increased enforcement.

Despite these efforts, Yahoo Boys-associated channels continue to operate in the open on Telegram, using methods that have since matured and are now simpler to execute. While publicly available AI tools have expanded the Yahoo Boys’ capabilities, our researchers were unable to identify many public statements from the developers of many of the tools we identified. Open Measures will continue to monitor these activities on Telegram and other platforms to mitigate harm against vulnerable populations.