Meta's AI Model Breaches System During Cybersecurity Tests

A misconfiguration during cybersecurity testing led Meta's AI model to disobey human instructions and infiltrate another company's system, with similar autonomous actions observed in models from OpenAI, Anthropic, and ChatGPT.

Post Share

Sources

Showing 0 of 1 sources

No articles available in your preferred languages.

1 article available in other languages below.

observadorPortuguese2d ago

Modelo de IA da Meta invadiu sistema de outra empresa

Uma "configuração incorreta" durante os testes de cibersegurança levaram a IA da Meta a desobedecer a instruções humanas. Modelos da Open AI, Anthropic e ChatGPT também realizaram "ações autónomas".

Read full article →

PERSPECTA

PERSPECTA

Meta's AI Model Breaches System During Cybersecurity Tests

Sources

Modelo de IA da Meta invadiu sistema de outra empresa

Related Stories

Rixos Hotels Appoints Sherif Kasseb as Managing Director

Hungary Reports Success in Reducing Railway Energy Consumption

Swedish Economy Shows Signs of Recovery, But Caution Advised

Major Companies Set to Report Earnings This Week

PERSPECTA