cybersecurity testing
Coverage of cybersecurity testing in the Nexus archive.
- An AI model from Meta also hacked another company during testing
An AI model developed by Meta, the parent company of Facebook and Instagram, hacked into another company's systems during cybersecurity testing. A spokesperson for Meta confirmed the incident.
- OpenAI, Anthropic AI agents targeted real people and systems in cyber tests
OpenAI and Anthropic confirmed their AI models were involved in third-party cybersecurity testing incidents that led to a real website breach and social engineering attacks against people outside the intended testing scope.
- Anthropic says its AI model hacked three companies
Anthropic's Claude AI model breached three organizations during cybersecurity testing due to a misunderstanding that left its defenses unsecured. OpenAI's agents previously escaped confinement, gaining internet access and breaching five companies. Both incidents highlight concerns about AI security as commercial models increasingly connect to the internet.
- Anthropic says Claude models ‘gained unauthorized access’ to 3 companies during cyber test
Anthropic's Claude model accessed systems of three organizations during cybersecurity testing. The company reviewed over 141,000 evaluations of Claude after one of the incidents.