AI-powered video surveillance expands from object detection to behavioral queries, raising privacy and misuse concerns

A Financial Times report describes how AI is changing video surveillance by enabling natural-language queries over video footage, allowing analysts to search for behaviors rather than relying on a limited set of preset object-based searches. This capability reportedly allows users to query footage for events such as two men handing a bag to each other, a person who has changed appearance or clothes multiple times in a day, or a vehicle that has been repainted or driven past the same location repeatedly in a short period.

Bruce Schneier, writing on his blog Schneier on Security, frames this development as part of a broader trend where AI enables mass spying, comparing it to how computers and networks enabled mass surveillance. He highlights that the new tools allow intelligence officers to ask open-ended natural language questions about video data and receive answers, rather than being constrained to a few dozen predefined search terms.

A European official quoted in the Financial Times characterizes the technology as "the holy grail of surveillance," stating that it creates "a world of new possibilities" by enabling searches based on behavior rather than objects.

Schneier and commenters on his blog raise concerns about the implications of such capabilities, including the potential for AI-generated evidence to be fabricated and submitted in courts. One commenter notes recent investigations have shown police submitting "AI-generated evidence" to courts and getting away with it, while another warns that "when truth becomes negotiable, even the cleverest inventions serve only the abyss."

The discussion also touches on broader societal impacts, with commenters arguing that ubiquitous surveillance—enabled by AI—could chill free expression and dissent, and that the technology's efficiency is higher in densely populated urban areas.