Delaware is developing a legal structure that would let AI agents manage companies without a human at the helm. A draft bill would create “Artificial Intelligence Companies” able to conduct nearly any lawful business except banking: agents could make transactions, sign contracts, manage property, and sue or be sued. A proposed 30-month regulatory sandbox would test the model under state supervision before lawmakers decide whether to establish it permanently. Supporters argue that giving autonomous systems a formal corporate identity would make their conduct traceable and provide a legal entity against which claims and damages could be pursued.
The plan’s central dispute is whether that structure would make AI activity accountable or instead shield its human backers. A company’s member would provide capital and receive profits, but generally would not be liable for the AI-run firm’s actions; exceptions include fraud, deliberate lawbreaking, inadequate capitalization, and knowingly allowing business outside the company’s stated purpose. House Oversight chair James Comer has asked Delaware Governor Matt Meyer for records, warning that the proposal could protect investors from harms caused by reckless AI and invite regulatory capture. Delaware officials have promoted the sandbox as a supervised experiment and a source of fees and business growth, while legal scholars have questioned the breadth of the liability shield and warned that states could compete to offer companies the most permissive rules. The effort involves AI firm Norm Ai and is being developed in a state where roughly two-thirds of Fortune 500 companies are incorporated.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.
Google AI Mode sometimes avoids directly saying Joe Biden won the 2020 US presidential election, instead directing users to official sources. Its responses vary across users and election years, while answers become more direct when asked outside the US. (arstechnica.com)
AI data centers can consume millions of gallons of water daily, competing with local communities for drinking water as their numbers grow. Residents around the world are organizing to oppose projects that threaten local water supplies. (restofworld.org)
Cartesia argues that tokenizers limit end-to-end learning because they impose hand-crafted text boundaries. Its hierarchical networks learn to group raw bytes into useful units, offering a scalable path toward byte-level language models. (cartesia.ai)
Particle-1 is a software-engineering model post-trained on DeepSeek V4.1 Flash that uses about 45% fewer tokens while maintaining comparable accuracy. Across four coding and data-engineering benchmarks, it cuts average task costs by 34.4%-51.6%, with plans to expand into other domains. (particle.ai)
Google’s Gemini gave different results for two election-related queries that differed by one digit, according to an X post. The post claims one result censored Joe Biden but does not show the queries or responses. (twitter.com)
Google introduced Gemini 4 Carbon, a new model in its Gemini lineup. The announcement describes the model and its capabilities. (blog.google)
Today's best Hacker News stories, summarized and screenshotted, one email a day.