Fake Anthropic AI invites lure employees into credential theft

Exploiting the AI hype cycle

Cybersecurity researchers from Cofense have identified a new phishing campaign targeting enterprise employees. The scam relies on the high demand for information regarding Anthropic, a prominent AI company. By offering a private, invitation-only preview of new AI systems, attackers tap into the professional curiosity of ambitious workers.

The lure is designed to feel exclusive and high-value. The emails describe a

Subtle clues in the sender details

While the phishing emails are designed to look professional, they contain minor inconsistencies that can serve as warning signs. For instance, the sender name appears as "Austin from nthropic," which is missing the first letter of the company name. While this is a small error, researchers suggest it is a useful detail for observant users to notice before interacting with the content.

The email sender address also uses a custom domain that mimics a legitimate service. The address appears to be customercare@gotowebinar[.]com, even though the actual website for that company is www.goto.com. These small discrepancies in branding and naming are often the only visible cracks in an otherwise convincing scam.

Using CAPTCHAs to block security tools

The attackers employ a clever technical strategy to prevent security software from detecting their malicious websites. They place Cloudflare CAPTCHA pages at multiple checkpoints during the registration process. Because CAPTCHAs are a common feature on legitimate websites, many people assume the site is safe and continue through the process without suspicion.

The primary goal of these CAPTCHAs is to keep out automated security scanners, VPN services, and Tor browsers. Since these tools are often used by security teams to analyze threats, the CAPTCHAs effectively hide the phishing site from them. This allows the human victims to proceed to the fake login pages while the security tools are blocked from seeing the content.

The burden of defense on leadership

Industry experts argue that the current culture of gated AI previews and waitlists makes these scams more effective. This business model trains employees to expect and seek out exclusive invitations, which hackers are now exploiting. Because the AI industry relies on these types of "private" events, users are conditioned to be more likely to click on a link that promises early access to new technology.

Security experts emphasize that the responsibility for preventing these breaches should not rest solely on an employee's ability to spot a fake URL at the end of a long workday. Instead, they suggest that protecting against stolen passwords is a leadership decision. Companies can implement phishing-resistant multi-factor authentication, such as passkeys, to ensure that even if a password is stolen, the attacker cannot easily gain access to the account.

The psychology of the lure

The success of this phishing campaign relies heavily on the psychological concept of exclusivity. According to Frank Dickson of Dickson Research, the scam works because it sells the idea of being part of an inner circle. By making the invitation feel like a rare opportunity, hackers exploit the fear of missing out that many professionals feel regarding new technology.

The specific language used in the emails is designed to be highly appealing to those in the tech space. The messages describe a private, invitation-only virtual session that brings together a select group of participants. These sessions supposedly focus on high-level topics like AI systems, research, and the future of frontier models, making the offer seem like a significant professional development.

Organizational defense strategies

Security experts suggest that companies should not rely solely on an employee's ability to spot a fake URL during a busy workday. While training is helpful, Dickson argues that the defense should not depend on a worker squinting at a link late on a Friday afternoon. Instead, the risk of a stolen password should be viewed as a leadership decision regarding how much the company is willing to let a single credential compromise its data.

To protect against these sophisticated scams, enterprises can follow a specific set of defensive steps. One primary recommendation is to verify any invitation for an AI preview through an official account team or a known contact rather than clicking links in unsolicited emails. Organizations should also treat any branding that changes across different pages as a major red flag and prioritize the rollout of phishing-resistant multi-factor authentication.