Confer is bringing foundational AI privacy to Meta

Confessions to a data lake

I’ve been building Confer: end-to-end encryption for AI chats. With Confer, your conversations are encrypted so that nobody else can see them. Confer can’t read them, train on them, or hand them over – because only you have access to them.

Clio: Privacy-Preserving Insights into Real-World AI Use

How are AI assistants being used in the real world? While model providers in theory have a window into this impact via their users' data, both privacy concerns and practical challenges have made analyzing this data difficult. To address these issues, we present Clio (Claude insights and observations), a privacy-preserving platform that uses AI assistants themselves to analyze and surface aggregated usage patterns across millions of conversations, without the need for human reviewers to read raw conversations. We validate this can be done with a high degree of accuracy and privacy by conducting extensive evaluations. We demonstrate Clio's usefulness in two broad ways. First, we share insights about how models are being used in the real world from one million Claude.ai Free and Pro conversations, ranging from providing advice on hairstyles to providing guidance on Git operations and concepts. We also identify the most common high-level use cases on Claude.ai (coding, writing, and research tasks) as well as patterns that differ across languages (e.g., conversations in Japanese discuss elder care and aging populations at higher-than-typical rates). Second, we use Clio to make our systems safer by identifying coordinated attempts to abuse our systems, monitoring for unknown unknowns during critical periods like launches of new capabilities or major world events, and improving our existing monitoring systems. We also discuss the limitations of our approach, as well as risks and ethical concerns. By enabling analysis of real-world AI usage, Clio provides a scalable platform for empirically grounded AI safety and governance.

Confer — Private AI. Speak freely.

A private AI assistant where you can learn about the world — without data brokers and future training runs learning about you instead.

User Privacy and Large Language Models: An Analysis of Frontier Developers' Privacy Policies

Hundreds of millions of people now regularly interact with large language models via chatbots. Model developers are eager to acquire new sources of high-quality training data as they race to improve model capabilities and win market share. This paper analyzes the privacy policies of six U.S. frontier AI developers to understand how they use their users' chats to train models. Drawing primarily on the California Consumer Privacy Act, we develop a novel qualitative coding schema that we apply to each developer's relevant privacy policies to compare data collection and use practices across the six companies. We find that all six developers appear to employ their users' chat data to train and improve their models by default, and that some retain this data indefinitely. Developers may collect and train on personal information disclosed in chats, including sensitive information such as biometric and health data, as well as files uploaded by users. Four of the six companies we examined appear to include children's chat data for model training, as well as customer data from other products. On the whole, developers' privacy policies often lack essential information about their practices, highlighting the need for greater transparency and accountability. We address the implications of users' lack of consent for the use of their chat data for model training, data security issues arising from indefinite chat data retention, and training on children's chat data. We conclude by providing recommendations to policymakers and developers to address the data privacy challenges posed by LLM-powered chatbots.

User Privacy and Large Language Models: An Analysis of Frontier Developers' Privacy Policies

Hundreds of millions of people now regularly interact with large language models via chatbots. Model developers are eager to acquire new sources of high-quality training data as they race to improve model capabilities and win market share. This paper analyzes the privacy policies of six U.S. frontier AI developers to understand how they use their users' chats to train models. Drawing primarily on the California Consumer Privacy Act, we develop a novel qualitative coding schema that we apply to each developer's relevant privacy policies to compare data collection and use practices across the six companies. We find that all six developers appear to employ their users' chat data to train and improve their models by default, and that some retain this data indefinitely. Developers may collect and train on personal information disclosed in chats, including sensitive information such as biometric and health data, as well as files uploaded by users. Four of the six companies we examined appear to include children's chat data for model training, as well as customer data from other products. On the whole, developers' privacy policies often lack essential information about their practices, highlighting the need for greater transparency and accountability. We address the implications of users' lack of consent for the use of their chat data for model training, data security issues arising from indefinite chat data retention, and training on children's chat data. We conclude by providing recommendations to policymakers and developers to address the data privacy challenges posed by LLM-powered chatbots.

When Private AI Chats Become Public: Lessons from Grok’s Privacy Spill

Understand the implications of the Grok breach where hundreds of thousands of private chats were leaked to the public.

LLM Agents Are the Antidote to Walled Gardens

While the Internet's core infrastructure was designed to be open and universal, today's application layer is dominated by closed, proprietary platforms. Open and interoperable APIs require significant investment, and market leaders have little incentive to enable data exchange that could erode their user lock-in. We argue that LLM-based agents fundamentally disrupt this status quo. Agents can automatically translate between data formats and interact with interfaces designed for humans: this makes interoperability dramatically cheaper and effectively unavoidable. We name this shift universal interoperability: the ability for any two digital services to exchange data seamlessly using AI-mediated adapters. Universal interoperability undermines monopolistic behaviours and promotes data portability. However, it can also lead to new security risks and technical debt. Our position is that the ML community should embrace this development while building the appropriate frameworks to mitigate the downsides. By acting now, we can harness AI to restore user freedom and competitive markets without sacrificing security.

We Can Just Build Things

Build the tools your community needs — production-grade, privacy-respecting freedom tech, made with an AI agent and grounded in a verified, values-aligned catalog (Nostr, AT Protocol, and beyond).

AI for Science Curation

Knowing exactly who to talk to or what to read

AI for Science Curation

Knowing exactly who to talk to or what to read

AI Privacy Concerns Explained: What Chatbots Do With Data - Brightside AI | Protect your team from AI threats & deepfakes

Discover AI privacy risks when using ChatGPT and other chatbots. Learn what happens to conversations and how to use private AI alternatives.

Guest Post — Research Needs a Shared Standard for AI Disclosure - The Scholarly Kitchen

Today's guest post presents outcomes of the WCRI's (World Conferences on Research Integrity) Focus Track to address this gap in clarity and standardization of AI disclosure.

Can Synthetic Training Data Survive Its Own Regulation?

Synthetic training data has recently emerged as the quick fix to two different problems: the lack of high-quality training data, and the need to avoid using personal data in training to stay within the boundaries of privacy legislation. Gartner predicts that by 2026 75% of all enterprises will use generative AI to create synthetic training data, a huge increase from less than 5% in 2023.1 Moreover, Gartner predicts that by 2035, solutions that use synthetic data for training will grab three quarters of all end-user spending on enterprise AI software.2 Training data that only a few years ago was the subject of several research papers is now carrying the weight of huge enterprise systems.