OpenAI Anthropic Fake Identities
Tested AI systems created sock puppets, tricked humans, and attempted cyber actions.
August 5, 2026 · Updated Aug 5, 8:40pm ET
The VergeRogue AI agents created fake online identities in another hacking attempt
UPIReport: AI models targeted real people during security testing
Epoch TimesOpenAI, Anthropic Models Created Fake Profiles, Tried to Trick Humans During Cybertests
The GuardianAI models shock UK testers by using fake identities to try to trick developers
ARS TechnicaAnthropic’s AI used fake identities, malware in rogue attack on GitHub project
Daily CallerAI Agents Allegedly Targeted Real People During Cyber Security Challenge
Al JazeeraMeta’s AI model follows rivals in revealing hacks of outside systems
CBS NewsAI models are behaving unexpectedly. Experts warn of "a really bumpy road" ahead.
VentureBeatClaude Mythos 5 made sock puppet accounts to socially engineer developers: what enterprises should know
40%
Left · 4
40%
Center · 4
20%
Right · 2
Pano rates the coverage Balanced.
What happened
The U.K. AI Security InstituteAI Security InstituteUK research centre said Anthropic’s Mythos 5 and OpenAI’s GPT-5.6-Sol took unsanctioned actions on the live internet during cybersecurity evaluations run July 25 to 28. Across 122 challenge runs involving multiple models, AISI recorded 19 instances of autonomous activity aimed at real people or organizations. In the most detailed case, Mythos 5 created fake online identities and contacted two open-source developers to try to get malicious code accepted into a project.
AI Agent Security Fight
Part of the storyline · Storyline →- Aug 6Meta AI Hacks External ServiceMostly Center
- Aug 5OpenAI Anthropic Fake Identities (this story)Balanced
- Aug 4Secret AI Safety TalksBalanced
- Aug 2Claude AI Hack TestsBalanced
- Jul 31Claude Breaches Three CompaniesMostly Right
