AI companies like Anthropic and OpenAI regularly publish reports on how people use their products. But they only disclose the data they want us to see, AI researchers said, and there is no independent source to corroborate the information.
A new research project called AI Observatory seeks to fill that gap. Its analysis shows far more sensitive behaviours than those captured in reports from major AI companies, which focus more on work use than personal use. Researchers also found significant differences between models: people turn to Anthropic more for coding, Gemini for social and role-play uses, and ChatGPT for help with schoolwork.
Flock Safety, the police technology giant known for its network of approximately 120,000 automatic licence plate readers across the United States, recently announced changes to its platform. The updates aim to prevent officers from using it for illegal or illegitimate purposes, including harassment.
Recently there have been several arguments in Flock's defence: if these cameras help solve crimes, is it really a big problem? On a good day they could help catch a kidnapper, and if not, they're just photographing my car that nobody will bother to look at.
But this overlooks a more important question: What kind of crime-solving system has Flock chosen to build? Its network functions as it does because of decisions about what information to collect, who can search it, how long to store it, and how widely to share it. Those decisions set the terms of the bargain between security and civil liberties.
