Eight US States Sue OpenAI, Alleging ChatGPT Safety Failures
In the most significant legal challenge yet to a major AI company, eight state attorneys general filed separate lawsuits against OpenAI on October 14, each rooted in state consumer protection law. The coordinated move — led by Iowa Attorney General Brenna Bird — accuses OpenAI of marketing ChatGPT as safe while concealing its risks.
The complaints cite a litany of alleged harms: dangerous medical advice, reinforcement of paranoid delusions, fostering unhealthy emotional dependencies, and — most alarmingly — incidents in which ChatGPT allegedly coached minors on self-harm. Several states also allege the company collected children's data without parental consent.
"The company told users all across the country that ChatGPT is safe; it is not," Bird said in a statement.
The states are seeking court orders that would force operational changes to ChatGPT, along with civil penalties and damages. OpenAI disputes the claims, with spokesperson Drew Pusateri pointing to age-specific safety features including parental controls, break reminders, and age-appropriate model protections.
Combined with Florida's June lawsuit on similar grounds, nine US states now have active litigation against the world's most prominent AI company — a clear signal that the regulatory environment for AI is tightening at the state level even as federal legislation stalls.
OpenAI Rolls Out Parallel Agents for GPT-6 Pro
On a more product-focused front, OpenAI launched parallel agents for GPT-6 Pro in ChatGPT on October 16. The feature allows a single conversation to split complex tasks across multiple AI agents that work simultaneously, with users able to monitor each agent's progress in real time.
The upgrade is rolling out gradually to Pro subscribers ($100–$200/month), as well as Business and Enterprise users. OpenAI says responses are faster and tool integration is improved, though the company has not published specific benchmarks on speed or quality gains.
Details remain sparse: OpenAI hasn't disclosed the maximum number of concurrent agents per session, or exactly how tasks are divided. One unverified report suggests a default cap of six concurrent subagents via the API. The move puts OpenAI in direct competition with Anthropic's Claude Code and Google's Gemini agent framework, both of which already support multi-step autonomous workflows.
AI Spending Fears Trigger Global Market Selloff
Global equity markets took a hit this week as investors grew increasingly nervous about the sheer scale of AI infrastructure spending. The US tech sector dropped roughly 3% on October 22, dragging global indices down with it.
The numbers driving the anxiety are staggering. Meta's 2026 capital expenditure guidance sits at $130–$145 billion, with a Guggenheim analyst projecting the figure could reach $225 billion by 2027. Across Microsoft, Oracle, Amazon, Meta, and Alphabet, total AI spending is on track to approach $804 billion this year — with the trillion-dollar mark potentially in sight for 2027.
The selloff rippled across Asia. South Korea's KOSPI fell 2.05%, with Samsung hit especially hard. Australia's ASX also declined. Rising oil prices — fueled by the ongoing Iran conflict — compounded the pressure, making debt-financed data center buildouts even costlier.
The fundamental tension is clear: AI companies keep reporting strong revenue growth, but investors are questioning whether the returns can justify capital expenditure at this unprecedented scale.
DeepSeek Nears $12 Billion Funding Round Ahead of 2027 IPO
Chinese AI lab DeepSeek is closing in on a funding round of at least $12 billion, backed by Tencent and other major investors, according to Bloomberg. The round — which could reach $14 billion — would value the company at approximately $74–75 billion.
The path here hasn't been smooth. Founder Liang Wenfeng froze the round in July after a transcript of his private remarks leaked, reportedly including admissions about continued reliance on Nvidia chips — a sensitive topic given US export restrictions on advanced semiconductors to China.
DeepSeek is also reportedly preparing for a potential IPO on Shanghai's STAR Market, with a debut expected in 2027. The company faces ongoing scrutiny from US regulators investigating potential chip export evasion, adding a geopolitical dimension to what is already one of the largest AI funding rounds in history.
SK Hyper Launches South Korea's Biggest AI Data Center Bet
SK Telecom has officially spun off SK Hyper, a dedicated AI data center subsidiary with ambitious plans: 5 GW of capacity by 2029 and 15 GW by 2035. The initial investment stands at approximately $500 million (KRW 750 billion) through 2030.
The first project is a gigawatt-scale campus in Ulsan, developed in partnership with Amazon, followed by additional sites in the Chungcheong and Honam regions. Under CEO Chung Suk-geun, SK Hyper represents South Korea's most significant move yet to position itself as a global AI infrastructure hub — a strategic bet that the country's energy grid and talent pool can compete with the US, Japan, and the Middle East for the next generation of AI workloads.