AI Legal Signal Brief

Delaware lawmakers have drafted a first-of-its-kind bill creating an “Artificial Intelligence Company” — a legal entity that can own assets, sign contracts, and be sued, while being run entirely by an AI agent with no human directly at the wheel. It's a 30-month sandbox experiment, not law yet, but it tells you state government already sees autonomous AI decision-making as a governance category that needs its own rules.

“Delaware Just Drafted a Company With No One Home.” Use it to reframe AI readiness as a fiduciary head start rather than a nice-to-have — the boards that can already answer “who's accountable when the AI decides” will be years ahead of the ones scrambling once this becomes real law.

If an AI tool inside your company made a call today that no human reviewed, who signed off on the limits it was operating under — and could you produce that record if someone asked?

A top-tier law firm is now telling boards directly that AI risk behaves differently from ordinary IT risk — it scales fast, changes after it's deployed, and depends heavily on third-party vendors the company doesn't control. That means a standard technology governance checklist won't catch it.

“3 Questions Your Board Should Be Asking About AI This Quarter” — pairs naturally with the message that most AI failures are a human and governance layer problem, not a model problem.

Can you name every place AI is actually being used in your company right now — not just the tools you approved, but the ones your teams picked up on their own?

A shareholder suit alleges Wix oversold how competitive and cost-effective its AI site-builder really was, with shares dropping 27% when the picture changed on an earnings call. It's the 7th “AI-washing” securities suit against an AI-developer company in 2026, and per Cornerstone Research, 2026's AI-related securities filings are on pace to nearly double 2025's.

“AI Washing Is the New Greenwashing.” Use the case as a study in the gap between public AI claims and what a company's own engineers would say privately — a gap plaintiffs' lawyers are now actively hunting for.

If an analyst compared your last earnings call to what your engineering team would say off the record about your AI capabilities, would they match?

Users who clicked “share” on a Claude conversation to send a link to a colleague ended up with that conversation indexed and searchable on Google — over 200 chats, some containing resumes, internal work details, and healthcare research. Nothing was hacked; a normal sharing feature simply behaved like a public webpage once a search engine crawled it.

“Your Employees Are Sharing More Than They Think” — a plain-language walkthrough of auditing default sharing and export settings across every AI tool in use company-wide. This is exactly the third-party tool and data-governance gap the NIST AI RMF “Govern” function is meant to catch.

Do you know what actually happens when someone on your team clicks “share” inside the AI tools you use at work — and who would find out first if it became public?