⚙️ Building AI rules | AI+ Government Summary

1. Key Themes

The U.S. Is Finally Building an AI Regulatory Framework — and Looking to Europe for Clues

The Trump administration, despite beginning with an anti-regulatory posture, is now constructing an AI oversight framework due by August 1. Europe's multi-year experience offers a concrete template.

"As AI models rapidly grew more powerful, the Trump administration has found itself developing an oversight framework and deciding exactly what should be covered."

The EU's approach centers on threat-scenario-specific benchmarking and independent verification — two methodological pillars the U.S. currently lacks.

"Other key lessons from the EU process include tying benchmarking — testing AI models against standardized evaluations to measure their capabilities and risks — to specific scenarios, such as bioweapons or government database hacks. The EU also emphasizes multiple, independent verifications rather than relying solely on companies' own assessments."

OpenAI and Anthropic Are Jointly Lobbying to Shape the Rules That Govern Them

In a notable industry coordination move, the two leading AI labs are pitching a unified framework to the administration — one that would cover all highly capable models regardless of how they're distributed.

"The Trump administration is weighing a unified pitch from OpenAI and Anthropic to create 'an equal playing field' where highly capable models are all covered, regardless of whether they're open- or closed-source."

Data Centers Are Becoming a Major Political Battleground — Pollution Is the New Front

The Democratic Party is evolving its data center attack from energy costs to environmental impact, eyeing the midterms and 2028. The key target demographic: Latinos.

"In a poll released this week by the firm, 64% of Latinos said keeping costs down and limiting pollution are equally important when considering how to meet data centers' energy demand."

Anthropic vs. the Pentagon: A Landmark Test Case for AI Company Autonomy

The legal standoff between Anthropic and the Defense Department over use-case limits for Claude sets a precedent for whether AI companies can enforce ethical guardrails against government clients.

"Anthropic insisted that two areas remain off-limits: mass surveillance of Americans and fully autonomous weapons."

The judiciary appears skeptical of the government's position, with the case weakening over time.

"I don't see additional evidence from the government really justifying what it did. If anything, it seems like the record, in some ways, has gotten worse for the government." — U.S. District Judge Rita Lin

2. Contrarian Perspectives

The U.S. anti-regulatory stance didn't prevent regulation — it just delayed it. The Trump administration's initial move to scrap Biden's AI executive order was framed as pro-innovation. But rapid capability growth forced a reversal, resulting in a framework being built under deadline pressure rather than deliberate design — potentially a worse outcome than iterative regulation.

"President Trump began his second term by scrapping his predecessor's AI strategy and pursuing an anti-regulatory regime. But as AI models rapidly grew more powerful, the Trump administration has found itself developing an oversight framework."

Government talent pipelines for AI safety are structurally broken. The brain drain from public to private sector isn't just a U.S. problem — it's actively hollowing out the EU and U.K. regulatory bodies that are supposed to be the global gold standard. This undermines the credibility of any framework produced.

"There have been several times where we're working with someone on the [U.K.] government side, and they suddenly put in their two weeks' notice, maybe less, and then they have a great position at OpenAI." — Chris Canal, EquiStamp

Clean-energy data centers may be a winning political issue — but Democrats haven't figured out the message yet. Despite strong polling data suggesting voter appetite, the party lacks a coherent communications strategy, limiting the political leverage of an otherwise powerful issue.

"But when it comes to messaging those goals effectively, the party is still figuring it out."

3. Companies Identified

Anthropic AI safety-focused lab behind Claude models Central to two stories: jointly lobbying for the U.S. AI framework alongside OpenAI, and fighting a landmark legal battle against Pentagon blacklisting.

"The Trump administration is weighing a unified pitch from OpenAI and Anthropic to create 'an equal playing field' where highly capable models are all covered." "Anthropic insisted that two areas remain off-limits: mass surveillance of Americans and fully autonomous weapons."

OpenAI Leading AI lab Co-lobbyist with Anthropic on U.S. framework design; also cited as a destination employer drawing talent away from U.K. government AI safety roles.

"The Trump administration is weighing a unified pitch from OpenAI and Anthropic to create 'an equal playing field.'" "They suddenly put in their two weeks' notice...and then they have a great position at OpenAI."

EquiStamp AI evaluation firm Case study in applied AI safety methodology; its CEO shaped EU and U.K. regulatory approaches and is now a key voice on best practices for the U.S.

"Researchers begin with historical data on threats such as biological weapons, then measure how much AI increases a bad actor's capabilities and whether that creates a systemic risk to society." — Chris Canal, CEO

Climate Power Climate-focused political communications firm Actively advising Democratic candidates to build voter messaging around data center pollution, with polling data targeting Latino voters specifically.

"Climate Power is advising campaigns that clean-powered data centers appeal to an unlikely coalition of liberals, college-educated women, voters of color and younger men."

Smartsheet (Sponsor) Enterprise work management platform Cited via sponsored content arguing that enterprise AI pilots fail due to fragmented systems, not model quality.

"95% of generative AI pilots produced no measurable business impact, according to MIT's Project NANDA."

4. People Identified

Chris Canal CEO and co-founder, EquiStamp AI evaluation expert who directly shaped EU and U.K. AI regulatory frameworks; primary expert voice in the EU lessons piece.

"Researchers begin with historical data on threats such as biological weapons, then measure how much AI increases a bad actor's capabilities and whether that creates a systemic risk to society." "There have been several times where we're working with someone on the [U.K.] government side, and they suddenly put in their two weeks' notice."

U.S. District Judge Rita Lin Federal judge overseeing Anthropic v. Pentagon Signaled strong skepticism of the government's case in successive hearings, with her rulings potentially reshaping AI companies' ability to set use-case limits.

"I don't see additional evidence from the government really justifying what it did. If anything, it seems like the record, in some ways, has gotten worse for the government." "I don't see evidence that Anthropic could alter the model after it was delivered or flip some kind of kill switch."

Marcus Robinson Spokesperson, Climate Power Driving the organization's campaign to arm Democratic candidates with data center messaging.

"Climate Power has been aggressively pushing out data and findings on voters' views of AI and data centers to campaigns increasingly looking to sharpen their message on energy affordability."

Rep. Alexandria Ocasio-Cortez (D-N.Y.) U.S. Representative Praised New York Gov. Kathy Hochul's data center moratorium and is sponsoring a national bill to pause new data center projects.

"New York Gov. Kathy Hochul's data center moratorium drew praise from Rep. Alexandria Ocasio-Cortez (D-N.Y.), who is behind a national data center moratorium bill that would go further and pause new projects nationwide."

5. Operating Insights

For AI companies operating in or near government: define your ethical hard limits in writing before contracts are signed. Anthropic's dispute with the Pentagon escalated precisely because use-case boundaries weren't agreed upon upfront. Establishing non-negotiable constraints before engagement — not after — is now a demonstrated legal and reputational necessity.

"Anthropic insisted that two areas remain off-limits: mass surveillance of Americans and fully autonomous weapons. The standoff escalated when the Pentagon designated Anthropic a national security and supply-chain risk."

For AI safety/evaluation firms: the EU methodology of scenario-specific benchmarking tied to historical threat data is becoming the reference standard. Companies building compliance, auditing, or red-teaming products should align their frameworks to this model rather than generic capability evaluations.

"Key lessons from the EU process include tying benchmarking...to specific scenarios, such as bioweapons or government database hacks. The EU also emphasizes multiple, independent verifications rather than relying solely on companies' own assessments."

6. Overlooked Insights

The EU's peer-reviewed science requirement for risk thresholds is a significant structural divergence from U.S. practice — and a potential competitive moat for evaluation firms. The EU grounds its AI risk thresholds in published, peer-reviewed science, while the U.S. relies on internal or industry-driven standards. This creates a meaningful gap in regulatory rigor that could affect cross-border AI deployment and give scientifically credentialed evaluation firms outsized influence in EU markets.

"The EU commits to getting buy-in from the scientific community, grounding AI risk thresholds in peer-reviewed science before embedding them into policy and pushing researchers to publish their methodologies. By comparison, the U.S. relies mainly on internal or industry-driven standards."

The centrist-progressive divide on data center moratoriums may be closing faster than expected. This intra-party schism has historically stalled Democratic AI legislation, but Hochul's moratorium receiving AOC's praise suggests the Overton window is shifting — which could have near-term implications for data center permitting and infrastructure investment timelines.

"The gap between centrists and progressives on the idea of data center moratoriums may be narrowing. New York Gov. Kathy Hochul's data center moratorium drew praise from Rep. Alexandria Ocasio-Cortez."