As AI drives rapid data-center development, states face growing questions about infrastructure costs and economic benefits.

To address these challenges, states should adopt a common, industry-informed framework for evaluating projects...🧵

Europe has put open-source AI at the center of its push for tech sovereignty and greater technological independence.

But while it is unlikely to match the United States and China in the race for the most powerful closed models, it has a real opportunity to lead in open AI...🧵

Opposition to data centers is growing—and increasingly taking on a partisan dimension.

But many of the concerns driving that opposition resemble those Democrats have confronted before with wind and solar projects: land use, noise, infrastructure costs, and local impacts...🧵

In @NRO, Daniel Castro (@castrotech) argues that as China races to build and deploy AI, U.S. policymakers shouldn’t let AI anxiety push America toward more government control and restrictions on innovation.

Doing so risks weakening one of America’s greatest advantages: its ven innovation...🧵

AI kill switches sound like a straightforward way to stop rogue AI systems.

But mandating them could introduce new cybersecurity and reliability risks—and still fail to address the broader challenge of containing autonomous AI...🧵

The House passed the Ratepayer Protection Act 417–3, sending the bill to the Senate.

But while the bill gets the core principle right—large electricity users should pay the grid costs they create—it shouldn’t single out data centers.

@techpolicymaven explains why the standard ge electricity users.👇
vist.ly/5ita3

Anthropic researcher @hilbertspaess resigned this month, warning that people building advanced AI believe it could “kill us all” by the end of the decade. Other AI leaders have called for slowing frontier development.

But frightening scenarios are not evidence that catastrophe is likely...🧵

In June, @CommerceGov temporarily blocked global access to @AnthropicAI's Fable 5 and Mythos 5 after concerns over a reported jailbreak.

Although they were restored weeks later, the episode exposed major uncertainty about how—and when—the U.S. can shut down frontier AI...🧵

Adversarial AI distillation—extracting capabilities from frontier AI models to train competing systems—threatens U.S. AI leadership & national security.

Congress is right to act, but the proposed Deterring American AI Model Theft Act of 2026 needs fixes before it becomes law...🧵

AI is transforming the workforce, & some universities are already adapting.

Institutions like @SyracuseU, @UofMaryland, @UW, & others are embedding AI across disciplines—not just computer science—to prepare students for an AI-enabled economy...🧵

.@AnthropicAI says fraudulent accounts generated 16M+ interactions to harvest @Claudeai outputs.

@OpenAI testified @DeepSeek_ai employees bypassed safeguards to extract model outputs.

These incidents put adversarial AI distillation squarely on policymakers' radar...🧵

Successful AI adoption depends on more than new tools. 🤔

Cities like Washington, DC, San Jose, Seattle, & Cleveland are showing that workforce training and AI literacy are essential to turning pilot projects into lasting improvements in public services...🧵

. @CNN is suing @perplexity_ai over its use of news content, but this case raises different legal and policy questions than many AI copyright lawsuits.

The key issue isn't just AI training—it's how AI search products use and deliver publisher content...🧵

The Great American AI Act discussion draft is a step toward a coherent national AI framework.

But Congress must get the details right—strengthening independent reviews, adopting dynamic compute thresholds, refining protections of people & preemption provisions, & expanding transparency requirements.

AI in state government is expanding through pilot programs, but most states still lack consistent evaluation and coordination—risking fragmented experimentation instead of scalable, evidence-based public service improvement...🧵

President Trump's AI innovation & cybersecurity executive order is a constructive step toward addressing AI-related cyber risks while preserving US AI leadership.

But as @TechPolicyMaven notes, policymakers must keep strengthening cyber defenses & resilience as it's implemented.

The federal CyberCorps Scholarship for Service program is being revamped to include AI-focused cybersecurity training as AI-powered cyber threats rapidly evolve

But training students alone won’t fix the government’s cyber workforce gap...🧵

White House officials are weighing whether advanced AI systems should be “proven safe” before release.

But requiring government pre-approval for frontier models would slow innovation, politicize deployment decisions, and weaken U.S. AI leadership...🧵

France is ordering ministries to phase out non-European tech providers—replacing tools from @Microsoft, @Google, @AWS, & @Zoom with domestic alternatives in a sweeping digital sovereignty push.

But at what cost to productivity?...🧵

False “Made in USA” claims are a real problem. President @realDonaldTrump's recent EO targets deceptive labeling—but its approach is flawed.

Shifting liability to online marketplaces misdirects enforcement & risks higher costs & fewer products for consumers...🧵

Sidewalk delivery robots are coming to U.S. cities.🤖🛣️

But most states are moving slowly. Policymakers should study China’s approach to safely scale this emerging technology...🧵

.@SenSanders & Rep. @AOC introduced a bill to halt new AI data center construction—but the case for a moratorium rests on a mix of misplaced concerns about AI safety, energy costs, & jobs.

See our full statement from @hodanomaar:
vist.ly/4w8fj

Agentic commerce is nearly here. 🛍️

AI agents will track prices, compare options, and make purchases on users’ behalf—saving time, lowering costs, and improving market efficiency.

But regulation hasn’t kept up...🧵

States are debating how to regulate AI in healthcare.

Utah offers a clear model: don’t block the technology—test it. Its regulatory sandbox lets AI tools operate under supervision, generating real-world evidence instead of relying on speculation...🧵

The @WhiteHouse has released its National Policy Framework for AI—offering Congress a clear, pro-innovation blueprint that targets real harms, supports adoption, & avoids a patchwork of state laws.

Our full statement from @castrotech:
vist.ly/4vk8n

The discussion draft of Sen. @MarshaBlackburn's TRUMP AMERICA AI Act aggregates a slew of heavy-handed, innovation-constraining ideas for regulating AI systems into a flawed federal framework.

.@GOVUK released its Copyright and AI report with no proposal or timeline.

While avoiding a flawed policy is better than rushing, continued inaction is unsustainable.

Our statement from @MRJKilcoyne:
vist.ly/4vctu

The @Europarl_EN has disabled built-in AI features on staff devices over security concerns, but bans don't reduce risk.

As @MRJKilcoyne explains, they sidestep EU cloud security frameworks, push AI into the shadows, and hurt productivity...🧵

The @DeptofWar–@AnthropicAI dispute shouldn't be settled in backroom negotiations.

In @techpolicypress, the Center's @castrotech explains Congress must step in to set clear statutory boundaries for military AI use.

🔗Read the piece:
vist.ly/4t5ca

The Center's Director @castrotech was in India last week for the #IndiaAIImpactSummit2026.

Here are his 3⃣ key takeaways from the summit for policymakers...

At the State of the Union, @POTUS announced a 'Ratepayer Protection Pledge' calling on major tech companies to build their own power plants for AI data centers — a pragmatic path that keeps costs off American households...🧵

AI-driven data center growth has raised concerns about household electricity costs. But the policy response matters.

@hodanomaar explains what the GRID Act gets wrong & how smarter market design can protect consumers while supporting innovation...🧵

EU leaders are calling for an “Airbus of payment systems” to reduce reliance on Visa and Mastercard.

But the EU’s payment sovereignty push would hurt consumers—addressing hypothetical risks, not real market failures...🧵

🚨 NEW: The EU often celebrates its ability to set global rules—a phenomenon known as the Brussels Effect. From #GDPR to the #DSA, EU regulations shape policies far beyond Europe’s borders.

But our latest report finds that this “regulatory export” comes with high costs for the Global South...🧵

At "Advancing Multilateral AI Partnerships," Panel 3 examined a shared roadmap for AI policy & innovation, with a focus on U.S.-India cooperation & governance alignment.

Featuring speakers from @google, @MahindraRise, @awscloud, & @USISPForum.

At "Advancing Multilateral AI Partnerships," Panel 2 examined compute equity & interoperable safeguards, with a focus on expanding access to compute while aligning on shared safety standards.

Featuring speakers from @ORFAmerica, @googlecloud, & @DataInnovation.

At "Advancing Multilateral AI Partnerships", Panel 1 explored how to scale AI adoption in practice—focusing on skills, deployment, trust, and real-world impact across sectors.

Featuring speakers from @CarnegieEndow, @ficci_india, & @CGDev.

Starting now in Washington, DC📍

Leaders from government, industry, & civil society convene to advance multilateral AI partnerships in a pre-summit event we're hosting with @_DialogueIndia ahead of the #IndiaAIImpactSummit2026.

President Trump's Proclamation invoking Section 232 of the Trade Expansion Act of 1962 imposes a 25% tariff on some foreign semiconductors.

ITIF VP Stephen Ezell notes: Build capacity first, don't escalate tariffs.

Our full statement👇
vist.ly/4n4we

What last year’s tech policy debates got right—and wrong—will shape 2026.⏪

One of the biggest fault lines?

Whether the U.S. should allow 50 different state AI rulebooks or set clear national rules.

Here's an explainer on federal preemption generated with @NotebookLM⬇️

New York claims its new AI safety law aligns with California’s framework, bringing the U.S. closer to a unified approach.

But as @hodanomaar explains, the law introduces rules that diverge sharply—creating a patchwork that threatens national coherence in AI regulation...🧵

Europe’s long-overdue ePrivacy reforms aim to ease cookie consent fatigue 💻

But their modest tweaks arrive too late and leave the outdated opt-in model—and the friction it creates—largely intact...

President Trump’s latest executive order on AI sends a clear message: the U.S. needs a cohesive national framework—not a patchwork of state laws—to stay competitive.

Read @castrotech's full statement on the order⤵️
datainnovation.org/2025/12/the-wh…

The @EU_Commission has opened an antitrust case arguing that Google’s use of AI in search & YouTube could be harmful to competition.

@castrotech explains why this risks freezing innovation rather than protecting it.

Statement:
vist.ly/4hzrb

The U.S. can’t lead in AI if every state writes its own rules.

Federal preemption creates a single, clear framework that protects consumers, supports innovation, and keeps America competitive.

@Castrotech answers 15 FAQs showing why common objections don’t hold up.👇
vist.ly/4fxyt

President Trump’s new Genesis Mission marks a major national effort to accelerate scientific discovery with AI.

The Center's Hodan Omaar highlights the unified push to speed breakthroughs that strengthen U.S. prosperity, security, and public benefit.

The EU’s new Digital Omnibus aims to simplify digital rules, but unless it confronts its flawed regulatory model, Europe will keep tweaking at the margins instead of enabling true digital innovation.

Read our full statement⤵️
vist.ly/4f76t

In response to calls for greater European Digital Sovereignty from French President @EmmanuelMacron & German Chancellor @_FriedrichMerz at today’s Franco-German Summit, Policy Analyst @MRJKilcoyne issues the following statement..

The AI-Related Job Impacts Clarity Act sounds like a step toward transparency—but its vague criteria could create confusion, mislead policymakers, and stigmatize AI adoption...

The EU shouldn’t block capable firms from competing in financial data services.

The success of the Financial Data Access (FiDA) framework depends on fair access, innovation, and consumer choice—not protectionism.

Today, the @EU_Commission unveiled the Apply AI Strategy to expand AI use in industry and government—but without fixing restrictive regulations, Europe risks falling further behind.

Read @castrotech's statement:
datainnovation.org/2025/10/brusse…