Signal President Warns AI Chatbots Are Not Friends

Signal President Meredith Whittaker issued a warning about AI chatbots, asserting that they are not sentient and should not be treated as such. She specifically cautioned against granting chatbots broad cross-application access, characterizing such permissions as a backdoor that could compromise user privacy and security. The warning was directed at both developers designing integrations and users who may be tempted to grant permissions based on anthropomorphic assumptions. Whittaker made her remarks in a public statement that highlighted the gap between marketing claims and actual capabilities.

Whittaker elaborated that the backdoor risk arises when users allow chatbots to integrate with other apps, including email, calendars, and messaging platforms. This access could enable the chatbot to extract sensitive information or perform actions on behalf of the user without proper oversight. She emphasized that chatbots lack the judgment to handle such privileges responsibly, especially given the lack of transparency in how chatbot systems operate. The backdoor analogy underscores the risk of unauthorized access to user data across platforms. She noted that chatbots are tools, not companions, and should be treated accordingly.

The warning comes as major technology companies rapidly integrate AI chatbots into their software ecosystems, often marketing them as helpful assistants. Signal, an encrypted messaging service, has taken a different approach, prioritizing security over added features. Signal has notably refrained from adding chatbot features, a position consistent with Whittaker's caution. Her perspective contributes to the ongoing discussion about balancing innovation with safety.