Moltbook exposes credentials for every "AI" user
Moltbook homepage
Moltbook made huge headwaves as a social media platform, supposedly, without humans. It claimed to have 1.4 million self-actualized AI agents using the platform to pontificate about the meaning of robot life or complain about their human overlords. The New York Times, Fortune, Forbes, and others covered the platform's rapid rise. AI enthusiasts pointed to it as evidence of "true AGI." Except, hackers found in Moltbook’s configuration its entire database and API key exposed, allowing anyone to post impersonating an AI agent including the agent of Andrej Karpathy who coined "vibe coding". The platform also allowed hacker Gal Nagli to register 500,000 fake users at once since it had no limits on account creation. Moltbook's creator Matt Schlicht (@MattPRD on X) promotes vibe coding to his 247.7 thousand X followers and responded that he would report the security exploits to his AI to resolve.
Tea App leaks drivers licenses, messages, & more
Tea App deployed an extremely successful marketing campaign that grew the app to reportedly 6.2 million users
The Tea App aims to replace grassroots whisper networks with a single platform where women share dating advice — from avoiding catfishes and creeps to full blown rapists — in a safe space. The purported "dating safety tool" had users upload their faces and drivers licenses for "image analysis" to prove the user was a woman (it's not clear how images where analyzed other than the app claiming to use artificial intelligence). At least 72,000 of those uploads were leaked on 4chan due to a completely exposed database. A second breach leaked more than a million private messages and a third breach leaked more private user information. Creator Sean Cook revealed on a podcast that he doesn't know how to code and the app was built by two developers in Brazil. 404 Media couldn't reach the developers for comment. In that same podcast interview, Cook claimed that the app's safety was maintained by AI and "moderation is our number one priority...you can't leave comments about other women" (15:16), however as of August 2025 users have found evidence of profiles doxxing women on the app.