Hackers Impersonate Google Forms To Phish User Login Information

Meta Platforms Technologies has secured US Patent 12,513,102 B2, titled “Simulation of a user of a social networking system using a language model,” granted on December 30, 2025.

The patent, originally filed in November 2023 by Meta’s Chief Technology Officer Andrew Bosworth, outlines a system that trains a large language model (LLM) on a user’s past social media activity including posts, comments, likes, chats, and voice messages to generate responses, likes, comments, or even simulated audio/video interactions on their behalf.

This AI “bot” activates when the user is unavailable, such as during a long break or after death, aiming to maintain engagement on platforms like Facebook and Instagram by mimicking the user’s style.

The technology starts with a pre-trained LLM, refined using user-specific data from interactions logged in Meta’s systems.

Users can grant permissions via an interface, choosing which data such as public comments versus private DMs feeds the model, ensuring some control over privacy.

Once deployed, the bot monitors newsfeeds for relevant content, crafts prompts with context (e.g., the poster’s relationship to the user or affinity score), and outputs actions such as “like” or comments that feel authentic.

For instance, it might respond differently to family posts versus friends, factoring in user profile details or event timing, such as birthdays.

Ethical and Legal Concerns

Meta’s spokesperson emphasized to Business Insider that the company has “no plans to move forward with this example,” noting patents protect ideas without guaranteeing implementation.

This follows stalled user growth on Facebook and prior AI “user” pilots that annoyed real people in late 2024.

Yet, the concept clashes with Meta’s memorialization policy, which freezes deceased accounts in their current state. Critics like University of Virginia sociologist Joseph Davis warn that it disrupts grief, creating confusion by simulating presence: “Let the dead be dead.”

Precedents highlight risks. Microsoft’s 2020 patent for deceased chatbots was deemed “disturbing” internally and shelved.

Amazon demoed Alexa mimicking a grandmother from a single minute of audio in 2022, but never launched it.[query] Startup 2Wai’s AI avatars drew “nightmare fuel” backlash, prompting a pivot.[query] Replika and You, Only Virtual emerged from grief, but face ethical scrutiny in “grief tech.”

From a cybersecurity perspective, this raises the risk of impersonation. Malicious actors could exploit similar LLMs for deepfake social engineering, phishing via “family member” accounts, or spreading misinformation post-death.

Weak permissions or data breaches could leak training data, enabling account hijacking. Defenses include robust verification for legacy access, AI disclosure mandates, and user-controlled kill switches.​

According to Malware Bytes, Meta’s innovation underscores AI’s dual edge: preserving digital legacies or eroding authenticity.

While business incentives like boosted engagement loom, societal unreadiness philosophical, emotional, regulatory likely shelve it. For now, users should review privacy settings and legacy contacts to safeguard their real afterlives online.

Follow us on Google News , LinkedIn and X to Get More Instant Updates. Set Cyberpress as a Preferred Source in Google.

The post Hackers Impersonate Google Forms To Phish User Login Information appeared first on Cyber Security News.