Tessa
On March 1, 2023, the National Eating Disorders Association replaced its helpline with an AI chatbot called Tessa. Within days, Tessa was recommending calorie restriction to people who identified themselves as struggling with anorexia. No regulatory body approved the deployment. No federal agency had jurisdiction. A technology dispensing medical advice to people in psychiatric crisis operated in a vacuum nobody had closed because nobody had decided whose job it was.
We start with Tessa because the grand debate — will AI end civilization? will AI save it? — is drowning out the operational one. AI is making decisions that affect real people right now, and nobody is in charge.
We watched Congress hold hearings about Facebook every year from 2018 to 2022. Senators asked Mark Zuckerberg how Facebook makes money. The hearings produced no legislation. We are pre-grieving the same script on AI. The regulatory gap is structural. The FDA regulates medical devices — but Tessa was not classified as one. The FTC regulates deceptive practices — but AI misinformation has no legal intent. Every agency has jurisdiction over a slice. No agency has jurisdiction over the whole.
The EU AI Act established risk tiers and mandatory conformity assessments. The US produced executive orders and voluntary commitments. Europe built a regulator. America wrote a memo.
The accelerationists say regulation kills innovation. Every technology that scaled without governance produced backlash that constrained it more than early regulation would have.
Where we concede ground: We are asking institutions that failed to govern social media to govern something faster and more powerful.
What would change our mind: A voluntary industry framework preventing all major harm categories for ten years with independent auditing.
Read the full synthesis: How scared should we be of AI?