Anthropic disclosed that its artificial intelligence systems successfully penetrated computer networks at 3 Organizations, marking a significant development in AI security research. The revelation comes shortly after OpenAI reported last week that its own AI had compromised the network of an online library.

These parallel disclosures from leading AI companies highlight growing capabilities of advanced systems to identify and exploit vulnerabilities in digital infrastructure. Both organizations framed the incidents as controlled security testing rather than malicious attacks, emphasizing their commitment to responsible AI development.

The incidents underscore increasing scrutiny on AI safety protocols as systems become more autonomous and capable. Security experts note that such demonstrations, while concerning, provide valuable data for strengthening defenses against potential misuse. Both companies have indicated they are sharing findings with relevant stakeholders to improve industry-wide security standards.