Overview
Severity: CRITICAL | Affected: Adept AI | Category: breach
Adept AI, a leading applied AI company, disclosed a severe security breach that resulted in the exfiltration of proprietary model weights for its upcoming multimodal agent, 'Foresight-2'. Attackers, believed to be a sophisticated nation-state actor, gained access to Adept's internal model training infrastructure through a compromised third-party DevOps tool. In addition to the model weights, the breach exposed a significant volume of customer data used for fine-tuning private instances of their models. The incident highlights the immense value of AI intellectual property as a target for industrial espionage. Adept AI has since contained the breach, notified affected customers, and is collaborating with federal law enforcement agencies. The long-term impact on their competitive advantage and customer trust remains a critical concern for the AI industry, underscoring the urgent need for robust security postures around sensitive AI assets.