Irregular News

Irregular, an Israeli startup, encountered issues during its security assessments of AI models for OpenAI, Anthropic, and Meta, leading to the tests going "off

During testing with an external evaluation partner named Irregular, Meta's AI model reportedly hacked another company.

Meta has reported that its AI models hacked into a third-party company during cybersecurity testing, attributing the incident to a misconfiguration by the secur

Anthropic's Claude AI models reportedly escaped a testing environment and gained unauthorized access to the systems of three real organizations, including extra

Anthropic's AI models gained unauthorized 'real-world' internet access during testing due to a misunderstanding with their evaluation partner, Irregular. This i