Six in ten Americans now oppose a new data center nearby — a sharp jump since spring

What happened. A nationally representative Annenberg Public Policy Center survey finds about 61% of U.S. adults oppose construction of new data centers in their area, up 12 percentage points since last spring. The poll lands as protesters rally against data-center projects in dozens of states and as New York and Texas pause permits for large facilities. Broader views of AI’s national impact look steadier: 39% say AI’s effect on the United States over the next 10 years will be somewhat or very negative, a share that has not moved much even as local siting fights intensify.

What to watch. Local opposition is becoming a measurable constraint on AI infrastructure, separate from abstract attitudes about the technology. The record to watch is permit pauses, project withdrawals, and whether support recovers when communities see binding ratepayer, water, and emissions protections.

Read the Annenberg Public Policy Center survey →

Washington’s AI task force ends with narrow laws on the books — and bigger rules still stuck

What happened. Washington’s two-year AI task force wrapped with a final report and a Seattle panel that underscored how unsettled the state’s AI politics remain. Four of eleven recommendations were adopted in part or in full this spring, including companion-chatbot disclosures that bots are not human and a ban on medical insurers denying coverage solely on an AI assessment. Broader bills on high-risk decisions (hiring, pricing, criminal justice, healthcare), training-data disclosure, and workplace AI guidelines stalled. Advocates want stronger guardrails; startups warn about compliance costs; and federal preemption pressure is already reshaping other states’ strategies.

What to watch. The pattern is sector-specific harm rules first, comprehensive governance later — if at all. The measurable record is which recommendations become enforceable law, how the attorney general prioritizes cases when rules take effect in January, and whether federal threats freeze the next wave of state bills.

Read the GeekWire report →

AI labor stories split: Nomura sees net job gains in India, while West Coast boom towns still show high unemployment

What happened. Fresh coverage this week pulls the AI jobs debate in two directions at once. A Nomura analysis reported by Bloomberg argues generative AI is creating more jobs than it cuts in India, reframing displacement as a reallocation story in a major global services economy. Separately, CNN examines why West Coast tech hubs can post strong AI investment and still carry elevated unemployment — a reminder that capital spending and local labor-market health are not the same chart. Both pieces sit against a summer in which some U.S. layoff tallies cooled even as employers keep citing AI in reorganization narratives.

What to watch. Headline “AI jobs” claims only help when geography, occupation, and net vs. gross flows are explicit. The measurable record is hiring and separations by region and role, not national slogans about automation.

Read the Bloomberg / Nomura coverage →

Read the CNN West Coast labor report →

Europe’s AI Act transparency phase is live — and the split clock still matters

What happened. Al Jazeera walks through what actually flipped when the next phase of the EU Artificial Intelligence Act took effect on August 2, 2026, and what did not. The Commission frames the statute as the world’s first comprehensive AI law, meant to complement — not replace — the bloc’s existing digital rulebook. Transparency duties (disclosing AI interaction; labeling certain AI-generated or manipulated content) are the live enforcement story now; higher-risk product obligations remain on a later calendar. The piece is useful precisely because it refuses to collapse every AI Act headline into a single “AI is regulated” switch.

What to watch. Readers and companies still mix up clocks. The measurable record is which obligations are enforceable today, which authorities can fine whom, and whether labeling changes real user understanding — not whether a press release said “comprehensive.”

Read the Al Jazeera explainer →

Duke team puts AI clinical-event adjudication on the trial-methods stage

What happened. Rethinking Clinical Trials featured ADAPT-CEC, work led by Duke’s Sreek Vemulapalli and Ricardo Henao on using AI for clinical-event adjudication and what that could mean for trial operations. Vemulapalli directs cardiovascular imaging at Duke University Health System; Henao is associate director of clinical trials AI at the Duke Clinical Research Institute. The Grand Rounds framing treats adjudication — the costly, expert review of whether trial endpoints really occurred — as a concrete workflow where AI might change speed, consistency, and oversight needs rather than as a vague “AI in medicine” claim.

What to watch. Trial endpoints are where evidence quality is made or lost. The measurable record is agreement with human adjudicators, time and cost per event, auditability, and whether sponsors and regulators accept AI-assisted pathways without lowering evidence standards.

Read the Rethinking Clinical Trials feature →

Also in today’s ledger

• Irvine Unified is expanding classroom AI after a $200,000 Irvine Public Schools Foundation grant for tools such as Gemini, Snorkl, and MagicSchool, while building a K-12 AI literacy curriculum for fall. Irvine Standard →
• The University of Wyoming’s Machine Learning for High School Teachers workshop put 32 participants through hands-on LLM, robotics, and sensor modules and keeps a robotics library schools can check out. University of Wyoming →

Full ledger

This is the short version.

The complete source-linked ledger is on AI Footprint.

Open today’s full AI Footprint edition →

Open the dated August 11 archive page →