Daily legal AI news
8 stories on AI in courts, ethics rules, regulation and legal software, each in a sentence or two with a link to the original source.
Attorneys general from eight states sued OpenAI on Wednesday, alleging ChatGPT was marketed as safe for children and other vulnerable users without adequate safeguards against harmful outputs.
Source: Arizona Attorney General (opens in a new tab) · Also: Bloomberg Law (opens in a new tab), Politico (opens in a new tab)
Forty attorneys general told a Delaware bankruptcy court Wednesday that Character Technologies Inc. should not sell Character.AI without enforceable child-safety and privacy protections.
Source: New York Attorney General (opens in a new tab) · Also: Colorado Attorney General (opens in a new tab)
California enacted Senate Bill 1000 on Wednesday, amending the AI Transparency Act to cover every generative AI provider with a publicly accessible system in the state, dropping the one-million monthly-user threshold.
Source: California Legislature bill status (opens in a new tab)
Thomson Reuters launched a Multinational Profile for CoCounsel Legal that it says researches across more than 100 countries and territories in one workflow.
Source: Thomson Reuters via PR Newswire (opens in a new tab)
Doe v. OpenAI Foundation, No. 3:26-cv-11781 (N.D. Cal., filed Oct. 13), alleges ChatGPT collected minors' data without effective parental consent and used it in training.
Anthropic's Guillaume Princen said open-weight models carry more cybersecurity risk "by definition" once public weights leave the developer's control and safeguards can be stripped.
Attorneys general who sued OpenAI Wednesday said more states may follow with similar ChatGPT safety cases.
England and Wales judiciary guidance still treats AI as support only: it cannot replace judicial reasoning, and new pilots need centralized approval.
Summaries are written by our team from the linked sources. They are news, not legal advice.