Use Deepfake Model API with Confidence: FriendlyAI Supports R1 with Serverless Endpoints - Digital Daily
Key Points
- 1Friendly AI has launched Korea's first service enabling secure API access to DeepSeek's high-performance R1 AI model, addressing widespread concerns over its data privacy policies.
- 2The "Friendly Serverless Endpoint" allows businesses to use R1 without personal data leakage by separating the model and applying Friendly AI's stringent security protocols, ensuring no user information is collected.
- 3This solution provides a more affordable, faster, and reliable alternative for AI model utilization, helping companies embrace AI transformation without privacy risks.
FriendlyAI has launched a novel service that enables the secure and privacy-compliant utilization of DeepSeek's high-performance inference AI model, R1, via an API in South Korea. This initiative addresses significant corporate concerns surrounding DeepSeek R1's data handling policies, specifically the perceived risk of excessive personal information collection and data transfer to Chinese entities, despite its strong performance comparable to OpenAI's o1.
The core methodology revolves around FriendlyAI's proprietary 'Friendly Serverless Endpoint'. This service, analogized to AWS Bedrock, operates on FriendlyAI's self-built cloud infrastructure, which integrates optimized model APIs. Unlike direct deployment of open-source models by companies on their private clouds to mitigate privacy risks, FriendlyAI's approach involves abstracting and securing the model access. The key technical aspect is the separation of the DeepSeek R1 model from the user's direct interaction. FriendlyAI acts as a secure intermediary, allowing third-party enterprises to call the R1 model's capabilities safely without direct exposure of their data or user information to the underlying DeepSeek infrastructure. This architectural design ensures that no user-specific data is collected by FriendlyAI itself, with the company explicitly stating that its established "strong standard security measures (Friendli Suite)" are rigorously applied to the DeepSeek R1 API, guaranteeing personal information protection.
From a user perspective, the service offers several key benefits:
- Enhanced Security and Privacy: By mediating access and separating the model, it alleviates concerns regarding data transfer and personal information collection associated with DeepSeek R1.
- Cost-Effectiveness: The service operates on a pay-as-you-go model, claimed to be more affordable and reasonable than direct R1 usage environments.
- Performance Preservation: FriendlyAI leverages its specialized AI model service expertise to maintain the original quality and performance of the DeepSeek R1 model while providing API access with improved response speeds.
- Ease of Integration: The serverless endpoint simplifies the integration and utilization of a powerful AI foundation model for enterprises.
FriendlyAI also highlights its recent success in providing services for low-cost, fast, and easy AI model deployment on Hugging Face, underscoring its broader capabilities in AI service acceleration and automation across various domains, including healthcare, telecommunications, home appliances, and content creation. The company's vision is to facilitate AI transformation for enterprises by demonstrating that high-quality generative AI services are achievable without incurring prohibitive costs.