Musinsa Tech said introducing the Model Context Protocol (MCP) for a conversational product discovery service made it easier for AI to connect to external data and software tools, but did not guarantee the AI would use tools as intended. The company said it faced trial and error in converting user questions into accurate search conditions and tightened input controls. It limited the service to non-login product discovery and blocked access to personal and payment data, stressing least-privilege design. It also called for wider native support and standardized authentication and security.

Binance runs an internal security programme that conducts monthly simulated phishing attacks on employees and can dismiss staff who repeatedly fail. Cointelegraph reported that Binance Chief Security Officer Jimmy Su said the results are reflected in employee performance evaluations. The drills are handled by Binance’s internal ethical hacking red team, which probes systems for weaknesses and designs fake attack scenarios. Staff who fail receive remedial training.