OpenAI tightened its internal security to protect against leaks after it alleged DeepSeek launched a new AI model by copying its models, FT reported. The security measures include limit on staff access to sensitive algorithms, isolating proprietary tech in offline computers and fingerprint access controls for office areas. OpenAI is also limiting internet access for external communications.