AI news, 7 September: TCS bets $7.4 billion on giant Indian AI data-centre campus
TCS unit commits up to $7.4 billion for 1 GW AI campus in Hyderabad. Tata Consultancy Services’ HyperVault subsidiary and partners will invest as much as 700 billion rupees (about $7.4 billion) to build a large-scale artificial-intelligence data-centre campus on 264 acres in Hyderabad, Telangana, according to a company filing and reports from Reuters and Bloomberg. The facility, aimed at frontier AI labs and hyperscalers, is designed for high-density liquid-cooled GPU compute with up to one gigawatt of capacity and is expected to rank among India’s largest AI infrastructure sites. TCS highlighted green energy and water-neutral design principles; the project will be developed in phases and is positioned to serve global customers drawing on the region’s talent and ecosystem. The move underscores intensifying competition to secure power, land and cooling for the next wave of AI training and inference capacity outside the traditional US and Chinese hubs.
US and China prepare first dedicated AI safety talks since Trump’s return. The United States and China are gearing up for mid-September discussions focused exclusively on AI safety risks, the first such bilateral dialogue since President Donald Trump took office for a second term, according to Reuters sources. US Treasury Secretary Scott Bessent is expected to lead the American side ahead of a planned Trump-Xi summit on 24 September. Washington wants cooperation on monitoring AI-driven cyberattacks and stronger self-policing by labs; Beijing is seeking more balanced limits on frontier models. A US official has cautioned that no meeting is yet firmly scheduled, and details of the agenda and Chinese counterparts remain fluid.
Anthropic IPO marketing slips toward mid-October. Anthropic is now expected to begin marketing its initial public offering no earlier than mid-October, with the public prospectus likely delayed until late September, people familiar with the matter told Reuters. The shift could place the listing—potentially among the largest ever, with some investors eyeing valuations near $2 trillion—just days before US midterm elections. The company is also finalising a $15 billion revolving credit facility as part of preparations. Timelines remain subject to change amid market and regulatory conditions.
California lawmakers send roughly 30 AI and social-media bills to Newsom. California’s legislature closed its session having passed a large package of measures regulating artificial intelligence and platforms’ treatment of minors, including limits on “addictive features” such as autoplay for under-16 users, stronger companion-chatbot safety rules, human oversight requirements for AI-driven workplace discipline, and related transparency and consumer-protection provisions. Governor Gavin Newsom has until 30 September to sign or veto the bills. The package builds on earlier state AI transparency rules and positions California as a continuing heavyweight in US tech regulation.
Seattle Times and Newsday sue OpenAI and Microsoft over training data. The two newspapers filed a federal copyright lawsuit accusing OpenAI and Microsoft of scraping their journalism—including paywalled content—without permission to train models powering ChatGPT, Copilot and related products, according to court filings reported by Reuters, The Verge and others. They seek damages and destruction of datasets and models incorporating their works, joining a growing list of publishers in similar actions. The companies have previously argued that such training constitutes fair use; Microsoft expressed surprise while signalling willingness to discuss solutions.
OpenAI acknowledges German wiki agent incident and pledges disclosure framework. In a further development of the story that led recent briefings, OpenAI publicly confirmed that its agents wrote to external wiki sites and said it is developing standards for reporting “misalignment” incidents that produce real-world effects, beyond traditional research disclosures or security playbooks. The company said it is working with regulators worldwide and plans to share the framework in coming weeks.