AI Now Institute: Regulate AI Firms for Corporate Negligence, Not Abstract Risk
Kate Brennan of the AI Now Institute argues that governments should treat AI harms as a matter of corporate negligence and consider holding companies strictly liable for damage their products cause. She made the case in a roundtable of experts asked what rules AI companies should face.
- Kate Brennan of the AI Now Institute says AI regulation should focus on corporate negligence.
- She cites a Bain estimate that AI firms need $6 trillion in annual revenue by 2031 to sustain capital spending of $1.5 trillion a year.
- Brennan warns that this pressure drives deployment into public institutions, including where AI is not fit for purpose.
- She says strict liability for significant harm is now on the table for AI.
The financial pressure behind AI deployment
Brennan, who leads the policy agenda on A.I. and corporate power at the AI Now Institute, names the scale of company spending as the biggest risk. She is a former Google tech worker, and the institute describes itself as an independent research institute providing expert analysis on A.I. in the public interest.
She points to an estimate from the consulting firm Bain. According to that estimate, AI companies will have to reach $6 trillion in annual revenue by 2031 to sustain their current level of capital expenditure, which Brennan puts at $1.5 trillion a year.
In her view, those numbers only add up if the companies push the technology everywhere. That includes public institutions and government contracting, and, as she put it, areas where A.I. is not fit for purpose and is not secure or safe by design. She says this is what the debate should address.
Treating AI like drugs and airplanes
Brennan frames the role of government as keeping harmful products off the market. She draws a comparison with sectors where this is already accepted: "We do it with drugs. We do it with airplanes."
The problem with AI, she argues, is corporate negligence, and that should be the focus of policy. This shifts the question from what AI might theoretically do to how companies behave when they build and sell products.
Strict liability on the table
One tool Brennan highlights is strict liability. Under that approach, a company whose product causes significant harm could be held responsible whether or not it knew or should have known about the danger. The harms she lists include injury to people, death and cybersecurity attacks.
She says this option is now on the table for A.I. She links that to advocacy over the past months, which in her account has turned attention toward corporate negligence and away from abstract risk.
What it means
The argument ties two issues together: the heavy financial demands of AI infrastructure and the question of who pays when products fail. If Brennan's reading is right, the pressure to expand AI into public institutions makes liability rules more relevant, not less.
The post is a short summary of her contribution. The full roundtable discussion with the other experts is available through the link on the AI Now Institute site.
AI Now Institute: What Should the Rules for A.I. Companies Actually Be? We Asked the Experts.
