Skill Leap AI - Watch This Before Using DeepSeek
The speaker highlights significant privacy issues with the Deep Seek website, emphasizing the lack of transparency in data collection and retention policies. The privacy policy indicates extensive data collection, including sensitive biometric data, without clear retention timelines or anonymization processes. Data is stored in China, raising compliance concerns with international privacy laws like GDPR. Users have limited control over their data, with no options to opt-out of data sharing with third parties. The speaker suggests alternatives, such as using a local version of Deep Seek R1 or accessing it through US-based services like Perplexity, to mitigate privacy risks.
Key Points:
- Deep Seek collects extensive user data, including biometric data, without clear retention policies.
- Data is stored in China, raising compliance issues with international privacy laws.
- Users cannot opt-out of data sharing with third parties, impacting privacy.
- Local installation of Deep Seek R1 or using US-based services can reduce privacy risks.
- Consider alternative AI providers for sensitive or critical data usage.
Details:
1. 🔍 Understanding Deep Seek's Privacy Policies
- Users are advised to thoroughly read Deep Seek's privacy policy and terms of use prior to using the website.
- There are three significant concerns within Deep Seek's privacy documents: data collection practices, user consent clarity, and third-party data sharing.
- A focused video provides a separate, in-depth examination of these privacy issues, distinct from regular content reviews.
2. 🚨 Key Privacy Concerns Unveiled
- The policy outlines various types of data collected, including profile data, user input, automatically collected data, and third-party data.
- Extensive user input collection involves text, audio, uploaded files, chat history, and feedback, but the policy does not clarify data retention duration or anonymization before analysis.
- There is no explicit mention of whether user data is used to train AI models or for model improvements.
- The policy mentions collecting keystroke patterns and rhythm, raising potential privacy concerns.
- Data retention and anonymization processes are not clearly defined, potentially impacting user trust.
- The policy lacks clarity on the use of user data for AI model development, which could affect transparency.
3. 🇨🇳 Data Handling and Storage in China
3.1. Data Handling and Storage Practices
3.2. Legal and International Implications
4. 🔒 Issues with Data Retention and Security
- China's data laws, such as the Cyber Security Law and Data Security Law, mandate companies to share user data with Chinese authorities upon request, potentially conflicting with international privacy standards.
- Companies with data centers in China, like DeepSeek, face risks as the government can request access to stored information at any time, challenging their ability to protect user privacy.
- For companies with users in the EU and the US, compliance with GDPR and other international privacy laws is a significant concern due to potential conflicts with China's data requirements.
- The lack of clarity on data processing compliance for EU users under GDPR highlights the need for transparent data management practices.
- Vague data retention policies, stating data is kept 'as long as necessary,' often fail to meet compliance standards, risking penalties and loss of trust.
- Companies like Facebook and Google have faced significant fines for GDPR non-compliance, emphasizing the importance of adhering to clear data retention timelines.
- Implementing more precise data retention guidelines and ensuring alignment with international laws can mitigate risks and enhance user trust.
5. 👶 Concerns Over Age Verification
5.1. Data Security Concerns
5.2. Age Verification Challenges
6. 📜 Legal Challenges in Terms of Use
- User data collection and sharing practices are a significant concern, particularly due to data storage in China and a lack of international transfer safeguards.
- The updated terms of use from January 20, 2025, continue to raise issues regarding data collection and usage, offering no explicit opt-out option for users.
- There is no clear indication of compliance with global privacy laws such as GDPR, and users are held fully responsible for AI-generated output.
- Dispute resolution is restricted to China, limiting users' legal recourse options, and it is unclear if users can request complete data deletion.
- Transparency regarding moderation decisions and data retention periods is lacking, suggesting data may be retained indefinitely.
- It is advisable to exercise caution when using the service for sensitive legal or business-critical purposes.
7. 🔄 Exploring Alternatives to Deep Seek
- Consider alternative AI providers that offer stronger privacy protection, legal compliance, and user rights safeguards than Deep Seek.
- One practical solution is to install a local version of Deep Seek R1, which is compatible with Mac, PC, and Linux and available on platforms like ama.com.
- While the online version of Deep Seek uses a 671 billion parameter model, this is impractical for local use due to its massive 400 GB size and high resource demands.
- Users can opt for a smaller, feasible option like the 70 billion parameter model, which can be run locally and has been tested successfully.
- Using Deep Seek locally allows users to bypass online terms and conditions, maintain privacy, and control their data by operating offline.
8. 💻 Local Installation Options for Privacy
- Local installation ensures full privacy by operating offline on Local Host 3000, without needing an internet connection.
- Users can download and switch between models like 32b, 671b, and 7B locally, although running the full 671b model may cause errors due to its size.
- For those unable to run the 671b model locally, Perplexity offers a Pro subscription with US-hosted model integration for $20, enhancing search with reasoning capabilities.
- Perplexity's R1 reasoning leverages information from multiple websites to deliver comprehensive answers, adding value to standard search functions.
9. 🛡️ Recommendations for Safe Usage
- Host the software locally in the US to address privacy concerns, ensuring better control and management of sensitive information.
- Avoid using the Deep Seek website for interactions involving sensitive data; opt for local alternatives or US-based partners to enhance data privacy.
- A planned comparison between ChatGPT's 01 model and Deep Seek's R1 model will assess their reasoning capabilities, aiding in selecting the most secure and effective tool.