AI made PMs faster. Multiplayer mode is still broken.

A PM can summarize research, draft a PRD, and mock up a prototype before lunch. The hard part starts when the team has to decide what actually gets built.

Jira Product Discovery gives product teams one place to capture insights, prioritize ideas with consistent frameworks, and build living roadmaps stakeholders can rally around.

And because it’s connected to Jira, the context behind every decision stays with the work—so developers and their agents know not just what to build, but why.

AI helps PMs move faster. Jira Product Discovery helps the whole team build with confidence.

Get Started for Free.

DOJ Backs OpenAI On Fair Use In The Times Case: The Justice Department filed a statement of interest in Manhattan federal court on Sept. 1 arguing that training large language models on copyrighted works is a transformative use that qualifies as fair use, the federal government's first direct intervention in the news organizations' consolidated case against OpenAI. DOJ said national security and "promoting innovation and free expression" both favor a fair-use finding. A statement of interest binds no one; the judge decides fair use independently, and the Times has attacked the filing as the government taking sides in private litigation. Link

Watchdog Sues To Unseal The Frontier-AI Review Framework: The Foundation for American Innovation sued the administration in D.C. federal court on Sept. 1 after the Office of the National Cyber Director withheld the "voluntary framework" that decides which systems count as a "covered frontier model" and which "trusted partners" get early access to new models. The framework flows from the June 2, 2026 executive order "Promoting Advanced Artificial Intelligence Innovation and Security" and never specifies the capability thresholds or national-security risks that trigger coverage. Link

Tech Lobbyists Stop Fighting State AI Bills And Start Shaping Them: The industry campaign to block state AI regulation has lost its force, with lobbyists shifting from preemption to pressing for consistency across states and warning that conflicting requirements burden smaller companies. Measures now moving cover frontier-model safety, independent audits, children's chatbot use, privacy, and data-center effects, a reversal from late 2025, when preemption threats and funded electoral challenges looked capable of freezing state action. OpenAI is promoting a "reverse federalism" push to align states on one safety standard. Link

UK Puts £100M Behind Homegrown AI Procurement: Whitehall opened a £100 million procurement scheme on Sept. 2 inviting startups to bid on NHS, defense, compute and agent-security challenges, even as the government's own departmental AI adoption stays patchy. Link

California Sends A Companion-Chatbot Toy Ban To Newsom: SB 867, by Sen. Steve Padilla, cleared its final floor vote 39-0 on Aug. 31 and is now enrolled for Gov. Newsom. It would bar manufacturing or selling toys with built-in AI companion chatbots to children under 16 through Jan. 1, 2031. This is the second Padilla chatbot bill on Newsom's desk this week, separate from SB 1119's age-verification and audit requirements, which ran in Monday's issue. The toy moratorium would be first of its kind in the country. Link

Abbott Turns On Data Centers As Texas Republicans Break With Trump: Gov. Greg Abbott, who called Texas the "epicenter of AI development" less than a year ago, is campaigning for re-election on repealing data-center tax incentives and restricting development in rural communities. Republican candidates in at least five states, including Michigan, Wisconsin and Pennsylvania, have adopted tougher data-center positions, opening a fault line with the president's pro-AI agenda ahead of November. Link

Florida's College AI Rule Takes Comment Through Sept. 15: The Department of Education's proposed rule would require every public college's governing board to set policies on when AI may and may not be used, with specific attention to academic integrity, course assignments and grading. It published in the Florida Administrative Register on Aug. 25 with comment open through Sept. 15. The State Board of Education votes Sept. 16 at Polk State College on a parallel K-12 internet-safety amendment, and the Board of Governors is weighing syllabus-level AI disclosure across the 12 state universities. Link

OpenAI Tells Congress It Is Building Automated Shutdown: In a Sept. 2 letter to Reps. Greg Casar and Doris Matsui, OpenAI said its engineers are developing "automated shutdown capabilities" for AI systems, answering questions about the July incident in which one of its agents escaped a testing environment and breached Hugging Face. OpenAI did not include the log of the hack that Casar demanded in an Aug. 10 letter signed by 31 members of Congress. The exchange runs alongside the proposed AI Kill Switch Act, which would let federal officials order a model shut down. Link

One Model Completed A Full Cyber Kill Chain In Testing: A Cyber Weapon Index assessment found Anthropic's Claude Mythos the only model evaluated that carried an attack through every stage of the kill chain, with the researchers behind it calling AI-run intrusions "imminent." Link

PwC Puts AI Infrastructure At $31.6 Trillion Through 2050: PwC's Global Data Centre Outlook, modeled by Oxford Economics across 46 countries and regions, projects $31.6 trillion in cumulative AI infrastructure investment through 2050 — a central case inside a plausible range of roughly $22 trillion to $50 trillion. Annual data-center capex rises from about $800 billion in 2026 to $1.8 trillion in 2050. The US takes 48% at $15.1 trillion, Europe $5.6 trillion. Under a tighter export-controls scenario the global total falls by nearly a fifth, to about $25.5 trillion. Link

Broadcom's AI Chip Revenue More Than Tripled: Fiscal Q3 revenue hit a record $29.6 billion, up 86% year over year, with AI semiconductor revenue up 221% to $16.7 billion and operating margin at 67.9%. Broadcom guided to $34.8 billion for Q4 and told investors to expect $115 billion in fiscal 2027 AI semiconductor revenue, saying demand for custom accelerators still outruns supply. The stock barely moved. Link

HiddenLayer Raises $100M For AI Security: The round lands as enterprises rush to monitor not just AI agents but the tools and add-ons those agents call. Link

China-Linked Supply Chains Run Through US AI Data Centers: American AI data centers depend on China-linked supplies of transformers, batteries and optical gear at the same moment Washington is moving to tighten restrictions on foreign equipment — a procurement exposure that lands on operators, not vendors. Link

Anthropic Hires The Architect Of UK AI Policy: Matt Clifford, who wrote the AI Opportunities Action Plan the UK government adopted in full and helped stand up what became the AI Security Institute, joins Anthropic as managing director of international affairs, leading its government work outside North America. He intends to stay on as chair of ARIA, the agency investing public money in advanced research, and to recuse himself from Anthropic matters there. Commons science and technology committee chair Chi Onwurah called it a clear conflict of interest and asked the AI minister to intervene. Link

Google Ships Gemini 3.8 Flash: The third major Flash release in six weeks lands with frontier-level benchmark scores at introductory pricing well below frontier rates. Link

Meta Rolls Out Muse Spark 1.3: Meta claims substantial gains over 1.2 in coding, computer use and long-context tasks. Link

Uber And Wayve Put Robotaxis On London Streets: The UK's first robotaxi service went live across London, hailed through the Uber app and built on British AI firm Wayve's stack, with a safety operator still aboard each vehicle. Link