Antidetect browsers add AI to run fake accounts at scale

Published

A single operator with a credit card and a prompt can now run thousands of social media accounts that each look like a different person, according to new research from Check Point Research. The firm examined three commercial antidetect browsers, Kameleo, AdsPower and MultiLogin, and found them referenced on Russian-language underground forums as building blocks for account takeover, carding, fraudulent loan applications and coordinated influence campaigns.

Antidetect browsers are legal, registered products sold for digital marketing, e-commerce and privacy. Their core trick is to give every browser profile its own fabricated fingerprint (canvas output, WebGL strings, fonts, time zone, screen size, user agent and hardware identifiers), so platforms that cross-reference those signals cannot tell that thousands of accounts share one machine. What has changed, Check Point argues, is AI: the manual labour that once capped how many accounts a team could run is being automated away.

An MCP server pipes AI posts into warmed-up accounts

The clearest example is AdsPower, which Check Point says advertises a dedicated Model Context Protocol (MCP) server, a standard interface that lets large language model tools drive other software. Paired with the product's RPA Marketplace of ready-made automation templates for liking, commenting, direct messages, follower scraping and AI-assisted engagement, it lets AI-generated content be produced, injected into pre-warmed profiles and pushed across platforms with no human in the loop. A companion Cookie Robot warms TikTok and X accounts by simulating browsing before they are put to work. AdsPower advertises a 99.2% ban avoidance rate, and its published pricing tiers support up to 5,000 profiles.

Check Point's view is that this delivers influence-operation scale and speed it previously associated only with state-sponsored actors. The report also notes that Meta's latest transparency reporting names Russia, Iran and China as the three leading sources of coordinated inauthentic behavior.

Stolen cookies become logged-in sessions

Kameleo modifies browser engines at the source level so its fingerprints come from real device signatures, suppresses WebRTC leaks of a device's true IP address, and offers a cookie injection API that accepts session IDs and refresh tokens. A session stolen by infostealer malware can therefore be loaded into a clean profile, and the target platform sees an already-authenticated user with no login event to flag. Check Point found a forum post pitching Kameleo as a component of Browser-in-the-Middle phishing kits aimed at Google, Microsoft, Facebook and financial accounts, and a recruitment post for an infostealer with RAT capabilities that lists Kameleo, AdsPower and MultiLogin among 21 compatible browsers. Another seller advertised an adversary-in-the-middle phishing engine that exports harvested sessions straight into antidetect browsers.

MultiLogin extends the model to mobile with cloud-hosted Android profiles, each with its own device parameters and proxy, plus partner virtual phone numbers for SMS verification. Check Point notes it also accepts cryptocurrency, SberPay and YooMoney, the latter two tied to Sberbank, which is under OFAC sanctions.

The report points to Anthropic's September 2026 threat intelligence report, which described a fraud campaign against cryptocurrency exchanges and financial marketplaces that combined antidetect profiles, residential proxies, automated CAPTCHA solving and identity-verification bypass tools to mass-produce verified accounts. Accounts like these pass KYC checks and build clean histories, so fraud systems that rely on behaviour or device identity often flag them only after the fraud has happened.

Treat a stolen cookie like a stolen password

For platforms and fraud teams, Check Point's findings mean that a consistent fingerprint and a residential IP address no longer prove a user is distinct. For everyone else, the cookie-injection path makes an infostealer infection on a staff or customer machine an account compromise in its own right, not just a password reset. We have covered the same mechanic when infostealers drained paid Claude accounts through stolen session cookies, and which secrets infostealers take most often.

The uncomfortable part is that none of this requires breaking anything. The tools are sold openly, explained in their vendors' own guides, and now wired to AI, which turns a team's worth of fake people into a subscription.

This briefing is provided by IntelFusions for informational and defensive purposes only. It is based on sources assessed to be reliable at the time of writing, and analytic judgments carry the confidence levels indicated. Indicators of compromise are defanged; re-arm them only in controlled environments. IntelFusions is not affiliated with the organizations named and makes no warranty as to completeness or accuracy.

Read the full analysis on IntelFusions