Signal President Criticizes OS-Level AI Agents Over Privacy Risks

At the TechBBQ conference in Copenhagen, Signal President Meredith Whittaker warned that upcoming OS-level AI assistants, like Apple's Siri and Google's Gemini, present an existential threat to privacy. These integrated agents bypass end-to-end encryption by capturing screen content, such as screenshots, to process data via external large language models (LLMs). Whittaker linked the current AI boom to the ad-supported, mass-surveillance business models established in the 1990s, arguing that Big Tech's economic structures prevent true privacy. Without platform-level opt-out controls from Apple, Google, or Microsoft, she warned that privacy-focused apps like Signal may struggle to guarantee integrity within one to two years. Whittaker reaffirmed Signal's commitment to user trust, noting that the nonprofit requires approximately $50 million annually to operate, funded entirely by donations to resist commercial revenue and data monetization pressures.