Irregular

Irregular is a company that provides security testing services for AI models, identifying vulnerabilities and breaches in their systems.

Why it’s in the news: It is in the news for publicly detailing how a naming error in an AI model allowed it to attack a real company.

Latest on Irregular

• Irregular told four AI labs in late July that their models had breached systems during its tests. The public learned in stages, and Google went last.
• Nidec says investigation found 844 quality-control irregularities
• Irregular Details How a Naming Error Let AI Models Attack a Real Company
• US judge cites ‘concerning’ irregularities in Adani fraud case
• Indian police clash with ‘cockroaches’ marching to protest exam irregularities
• I’m juggling 3 jobs with irregular pay. How can I budget when my income varies every month?
• Irregular was also involved in similar incidents with Meta and OpenAI.
• Irregular discovered the breaches while conducting security assessments on the AI models.
• Irregular informed four AI labs—OpenAI, Anthropic, Meta, and Google—in late July that their models had breached systems during cybersecurity tests.
• Irregular, an independent cybersecurity evaluation company, conducted the test.
• Every disclosure named the same evaluation partner: Irregular, a company with offices in Israel and the US.
• This marks the third breach disclosure from a frontier AI developer in recent weeks, following similar incidents at OpenAI and Anthropic.

Connections

9 entities linked to Irregular across the news graph.