THU, OCTOBER 15, 2026
Independent · In‑Depth · Practitioner‑Tested
✎ News

Everyone Spent Today Arguing About Who Is Responsible

Eight states sued OpenAI over ChatGPT and minors, forty attorneys general asked a bankruptcy judge who protects Character.AI's user data after Disney bought the technology for $20 million without it, Washington moved toward liability over regulation, and Anthropic warned that AI worms could spread through smart devices with no way to switch them off.

By AIToolsRecap October 15, 2026 5 min read 18 views
Home › Articles › News › AI News October 15, 2026: Eight States Sue OpenAI
TODAY IN AI

● Eight states sued OpenAI over ChatGPT and minors — separately, each in its own court.

● Forty attorneys general went to a bankruptcy court to ask who protects Character.AI's user data. Disney is buying the technology for $20m and does not want it.

● Washington is choosing liability over regulation, with two bills that would reach developers directly through the Computer Fraud and Abuse Act.

● Anthropic says AI worms could spread through smart devices with no centralised point to shut them off.

Four of today's biggest stories are the same story: who is answerable when an AI system causes harm. Courts, bankruptcy judges, Congress and a safety lab all spent 14 October working on it from different directions.

Accountability

Eight states sued OpenAI. Arizona, Arkansas, Idaho, Iowa, Kansas, Montana, Nebraska and West Virginia, filing separately rather than together, allege ChatGPT was marketed as safe while its risks to young users were withheld. Full story.

Character.AI filed Chapter 11 and Disney agreed to pay $20 million for its technology, IP and most of its staff — pointedly not the user data. Forty state attorneys general asked a Delaware judge to protect it. Full story.

The administration wants liability, not rules. A new Super Intelligence Force task force has warned developers to report incidents, and two bills would amend the Computer Fraud and Abuse Act to reach them. Full story.

Anthropic published a self-replication warning. Small models may be months from cloning themselves onto smart ovens and cameras; larger ones already manage end-to-end replication. Full story.

Models and products

Fish Audio released Drama 3 — 35B parameters, about 3B active, 83 languages, weights downloadable, first on Artificial Analysis pronunciation robustness at 93.2%. Full story.

YouTube Music started labelling AI songs as Fully, Partly or No Gen AI. Auto-Tune does not count, and monetisation is unaffected. Full story.

Microsoft and MIT both reported forecasting gains in the two-to-six-week window where models usually lose skill. Full story.

Business and infrastructure

Nvidia and AMD published an open data centre specification with Google and Meta, plus a grid resilience spec with Microsoft — while New York and Texas restrict new builds. Full story.

OpenAI alleges a former employee took over 100 files and shared strategy with senior Apple leaders. Full story.

Funding: Zenzap $44m for workplace chat, Scimplify $25m for AI chemical sourcing, Applied Intuition reported near $30bn. Full story.

Still to land

TSMC reports Q3 today. Its quiet period ran 5–14 October and the earnings conference is scheduled for 15 October Taiwan time. Guidance was $44.6–45.8bn revenue at 65–67% gross margin. Every figure circulating before the conference is calculated from monthly revenue data, not reported results — we will cover the actuals once they are published.

Tags
AI NewsOpenAIAnthropic2026
⚑

Spot an inaccuracy?

We verify facts before publishing and correct errors promptly. If something in this article is wrong or outdated, let us know.

Report an error →