June 21, 2023 MDG

Privacy preservation tactics using AI services

Best practices and methods

  • Use local models first

    Using AI at a local level essentially means running AI models on your own hardware, rather than relying on cloud-based services. This approach enhances privacy because your data doesn’t need to be sent to external servers for processing. Here are some top ways to access AI locally:

    Local Large Language Models (LLMs): Local LLMs are versions of language models that can run on personal computers. These models are typically smaller and less capable than their cloud-based counterparts, but they offer enhanced privacy.

    Example LLaMA.cpp is an implementation that allows running LLaMA models on consumer hardware. Users can download the model and run it entirely on their own machine.

    If  you are using services like OpenAI and others here are some basic best practices

  • Avoid sharing sensitive personal information on platforms that extensively use generative AI.
  • Use generic identifiers or pseudonyms instead of real personal details when interacting with generative AI models.
  • Report privacy concerns or suspected breaches of private data to the service provider and relevant regulatory authorities.
  • Utilize a Virtual Private Network (VPN) to anonymize user traffic, hiding location and preventing AI tracking.
  • Read and understand the privacy policies of generative AI platforms, ensuring transparency and data protection practices.  Often these are MIA. There is a reason for that.
  • Clear chat history or delete stored conversations to minimize data retention and analysis.  Request a copy first and save it.  Recommend this weekly or daily depending on your usage.
  • Use plug-in’s sparingly as there is no clear understanding of how privacy is preserved across various plug-in providers
  • Don’t use AI browser extensions that have full access to your internet activity.  Only install what you *must* install.
  • Verify if the generative AI service applies encryption and uses secure connections for data transmission.
  • Prioritize well-known and reputable generative AI platforms with a strong commitment to privacy and data protection.
  • Explore privacy-focused tools and browser extensions that block tracking scripts and enhance online privacy.
  • Maintain strong passwords, regularly update software and applications, and consider using a VPN for enhanced privacy protection.
, , ,