ChatGPT: weekly briefing
18 Oct – 25 Oct 2026
Top stories
- Texas sues OpenAI and Sam Altman over ChatGPT safety claims (Courthouse News, 23 Oct) — Texas Attorney General Ken Paxton filed a 67-page consumer-protection suit in Collin County, alleging OpenAI misled consumers and lacked adequate safeguards for teenage users. The state seeks a broad injunction; neither OpenAI nor the attorney general's office had responded to requests for comment.
- Iowa and Arkansas sue OpenAI over ChatGPT, reports say (Becker's Hospital Review, 20 Oct) — Becker's Hospital Review reported that Iowa has sued OpenAI over what it describes as dangerous medical advice from ChatGPT. Arkansas's attorney general has also reportedly sued over ChatGPT safety claims; the material does not confirm a wider count of states.
- OpenAI resumes frontier model training after security overhaul (The Standard, 24 Oct) — OpenAI said it restarted training of its most capable models on 1 October, after a September configuration error let a research model gain limited internet access. It also disclosed a flaw in the Brotli compression tool found during its testing.
- Sweden blocks ChatGPT and Claude browser access on government computers (CEOWORLD Magazine, 24 Oct) — Sweden's Government Offices will block browser versions of ChatGPT and Claude on staff computers from 26 October, following a security review, and approve Microsoft Copilot. Staff can still reach both models through an internal governed platform, Intric.
- OpenAI reportedly nears $30bn raise at $1.25tn valuation (The Motley Fool, 24 Oct) — The Motley Fool reported that OpenAI is about to raise $30 billion at a $1.25 trillion valuation, below the $1.5 trillion it had initially sought. Reuters reported the company is seeking to speed up fundraising ahead of Anthropic's planned IPO.
Trends
Several stories point to growing scrutiny of OpenAI's products and governance. Texas's lawsuit centres on teenage safety and the Hugging Face breach, and Iowa and Arkansas have reportedly filed their own actions, suggesting a multi-state pattern. OpenAI's pause and security overhaul of frontier training, prompted by an AI-agent incident, reflects similar concern about system safeguards. Sweden's browser block on ChatGPT and Claude shows public bodies applying comparable caution about data and reliability, though it is a government procurement decision rather than a legal action. The fundraising report and Uber's enterprise rollout stand apart as commercial news about growth and business demand.
What it means (opinion)
Opinion: The week's stories suggest that OpenAI faces pressure on several fronts at once: state attorneys general, public-sector purchasers and its own security reviews. Whether the Texas suit and reported state actions gain momentum will depend on the court's handling of the injunction request and on whether other states follow Iowa and Arkansas. Sweden's approach indicates that governments may favour contracted, governed access over consumer tools, which could shape procurement well beyond Europe. The fundraising at a lower valuation than sought, alongside the Anthropic IPO, suggests investors are more selective. Next week, watch for OpenAI's response to the Texas filing, any further state suits, and whether the security disclosures are independently verified, since safety evaluators said they had not yet checked the patches.
AI-assisted summary of public reporting; the analysis is opinion. Sources are linked above.