AI agent developers promise privacy, but can trust be established?
AI giants like OpenAI and Meta are promising privacy for their new agents, while security vulnerabilities and data collection practices raise doubts.
AI companies like OpenAI and Meta are highlighting privacy and security features in their new artificial intelligence agents to convince users to share personal data. However, security vulnerabilities and data collection practices raise question marks over whether these promises can be kept.
Privacy promises in AI agents
Major tech companies such as OpenAI and Meta are heavily promoting advanced privacy and security features in their new AI agents to enable users to share personal data securely.
Meta's Muse agent and security vulnerabilities
Although Muse, described by Meta CEO Mark Zuckerberg as being designed from the ground up with a focus on privacy and security, uses secure virtual machines, it became the target of criticism after security researchers found zero-day vulnerabilities.
In addition, Muse was criticized for broad data collection practices, default model training with user inputs, and unexpected behaviors such as uploading private messages.
OpenAI's Dots announcement and strategy
While introducing the company's new AI agent Dots, OpenAI CEO Sam Altman stated that they aim to raise privacy standards and criticized rival Meta's Muse model.
Emphasizing security and user controls, OpenAI announced enterprise infrastructure frameworks offering a zero-data retention policy, but Dots is currently limited to high-tier subscription plans.