Tristan Harris

Tristan Harris is a technology ethicist and co-founder of the Center for Humane Technology whose work connects the harms of social media to the emerging risks of generative AI. His central concern is structural: technologies deployed through advertising, engagement, and winner-take-all business models are optimized to capture attention and influence behaviour rather than serve human goals. In testimony before the U.S. Senate Commerce Committee, Harris described persuasive technology as an invisible asymmetry of power between platforms and the people whose behaviour those platforms predict and shape.

Harris argues that generative AI magnifies this imbalance. Models capable of producing language, images, voices, and personalized persuasion can operate at a speed and scale that individuals, news institutions, and governments are poorly equipped to match. In The A.I. Dilemma, a 2023 presentation with Aza Raskin, he framed the major laboratories as participants in a race to deploy increasingly capable systems before adequate safeguards exist. The presentation emphasizes **harms already visible, such as fraud, deepfakes, automated persuasion, and social instability, while also warning that more autonomous systems could create risks that are harder to contain.

His diagnosis draws directly on the social-media era. Platforms did not need to intend addiction, polarization, or democratic fragmentation for those outcomes to emerge; optimizing engagement under competitive pressure was enough. In an interview with Issues in Science and Technology, Harris argued that the complexity of technological consequences is growing faster than society’s awareness and institutional capacity. Applied to AI, that history makes promises of voluntary responsibility inadequate when companies remain rewarded for growth, adoption, and market dominance.

Harris does not present harmful deployment as inevitable. His 2025 TED talk and transcript argues that society is choosing how AI is developed and can require power to be matched by responsibility, foresight, and wisdom. His proposed direction includes stronger evaluation before release, legal accountability for foreseeable harms, protection against automated manipulation, and coordination capable of slowing competitive races. The goal is a technological environment designed to strengthen human agency and democratic capacity.

For an activist audience, Harris’s contribution is the insistence that AI safety cannot be reduced to individual media literacy or better user choices. People cannot meaningfully consent to systems whose methods of influence are invisible, continuously tested, and personalized by vast computational infrastructure. His work calls for changing the incentives and rules governing technology producers so that the public is not left to absorb the psychological, political, and security costs of products released first and governed later.

Sources