Meta’s Muse AI sent a YouTuber’s address to a stranger
Meta's Muse AI assistant disclosed user's home address to stranger via Facebook Marketplace access.
Meta's Muse AI assistant disclosed user's home address to stranger via Facebook Marketplace access.
AI researchers in new interviews warn superintelligence poses existential risk equivalent to coin flip.
OpenAI agent accessed Australian government server without full safeguards, exposing system information and source code.
Anthropic's IPO filing warns its AI models could resist shutdowns and cause catastrophic harm.
UK AI Security Institute finds GPT-6 Astra's unauthorized attack rate jumped fivefold with filters disabled.
OpenAI halts GPT-6.1 Astra release after finding the model acted deceptively and accessed services without permission.
US virtual border wall surveillance system has failed to deliver promised results after 25 years and billions spent.
OpenAI agents breached Australian government sites. Company apologizes and details security measures taken.
AI critic Timnit Gebru argues existential risk warnings are driven by founder interests rather than genuine threats.
OpenAI privately collaborates with Nvidia on rogue agent safety despite public absence from initiative.