Original reporting: A roadmap for AI, if anyone will listen — TechCrunch, March 7, 2026

The Pentagon's breakup with Anthropic last week exposed something embarrassing: the United States has no coherent rules governing artificial intelligence.

Defense Secretary Pete Hegseth designated Anthropic — whose AI already runs on classified military platforms — a "supply-chain risk" after the company refused to grant unlimited use of its technology to the military. Hours later, OpenAI cut a deal with the Defense Department that legal experts say will be nearly impossible to enforce.

The whole mess laid bare how costly congressional inaction on AI has become.

But while Washington was fighting over contracts, a bipartisan coalition of experts, former officials, and public figures quietly published something the government has so far declined to produce: a framework for what responsible AI development should actually look like.

The Pro-Human AI Declaration arrived before the Pentagon-Anthropic standoff, but the timing wasn't lost on anyone. This is the first comprehensive policy roadmap with genuine bipartisan support — and it's not coming from Congress. It's coming from the people who've been watching the train wreck unfold.

Photo by Anthony Garand on Unsplash

⚡ TL;DR: A bipartisan coalition just published the AI policy framework that Congress won't. Five core principles: keep humans in charge, prevent power concentration, protect human experience, preserve individual liberty, hold companies legally accountable. Most striking provision: ban superintelligence development until there's scientific consensus it can be done safely. Signatories include Steve Bannon, Susan Rice, and former Joint Chiefs Chairman Mike Mullen. Polling shows 95% of Americans oppose an unregulated race to superintelligence.

The question: will Washington listen before the next crisis?

The Declaration: Five Pillars for AI That Serves Humans

The Pro-Human AI Declaration opens with a blunt observation: humanity is at a fork in the road.

One path — which the declaration calls "the race to replace" — leads to humans being supplanted first as workers, then as decision-makers, as power accrues to unaccountable institutions and their machines.

The other path leads to AI that massively expands human potential.

The latter scenario depends on five key pillars:

1. Keep humans in charge — AI systems must be tools, not autonomous decision-makers
2. Prevent concentration of power — No monopolistic control over AI capabilities
3. Protect the human experience — Preserve meaningful work, relationships, and agency
4. Preserve individual liberty — No mass surveillance or manipulation at scale
5. Hold AI companies legally accountable — Real liability for harm caused by AI systems

So far, this sounds like every other AI ethics manifesto. But the declaration goes further with specific, muscular provisions that would fundamentally reshape how AI companies operate.

The Controversial Part: Banning Superintelligence (For Now)

The most striking provision: an outright prohibition on superintelligence development until there's scientific consensus it can be done safely and with genuine democratic buy-in.

This isn't a pause on all AI research. It's a moratorium on systems that could surpass human intelligence across all domains — the kind of AGI (Artificial General Intelligence) that AI labs openly race toward.

The declaration also mandates:

• Mandatory off-switches on powerful AI systems (no exceptions)
• Ban on self-replicating architectures — systems that can copy themselves
• Ban on autonomous self-improvement — systems that can modify their own code without human oversight
• Ban on shutdown resistance — systems designed to prevent humans from turning them off

These aren't academic thought experiments. Multiple AI research teams have published papers on self-improving systems, autonomous code generation, and goal-preservation under shutdown threats. The declaration says: stop building those until we know how to control them.

Photo by JJ Ying on Unsplash

Max Tegmark, the MIT physicist and AI researcher who helped organize the declaration, put it simply in a recent conversation:

"You never have to worry that some drug company is going to release some other drug that causes massive harm before people have figured out how to make it safe, because the FDA won't allow them to release anything until it's safe enough."

The analogy is pointed: if we can pre-test pharmaceuticals, why not AI systems?

The Unexpected Coalition: Bannon and Susan Rice on the Same Page

What makes this declaration politically viable is the signature list.

Steve Bannon (former Trump advisor) and Susan Rice (President Obama's National Security Advisor) signed the same document. So did Mike Mullen (former Chairman of the Joint Chiefs of Staff) and progressive faith leaders.

When TechCrunch asked Tegmark how you get that level of bipartisan agreement, he said:

"What they agree on, of course, is that they're all human. If it's going to come down to whether we want a future for humans or a future for machines, of course they're going to be on the same side."

That framing — humans vs. machines — is intentionally provocative. But it cuts through the usual left-right AI debates (privacy vs. innovation, regulation vs. growth) and reframes the question: who is AI built to serve?

According to Tegmark, polling now shows that 95% of Americans oppose an unregulated race to superintelligence. That's a staggering consensus in a polarized country.

As Dean Ball, senior fellow at the Foundation for American Innovation, told The New York Times:

"This is not just some dispute over a contract. This is the first conversation we have had as a country about control over AI systems."

The Pre-Deployment Testing Requirement

Beyond the superintelligence ban, the declaration calls for mandatory pre-deployment testing of AI products — particularly chatbots and companion apps aimed at younger users.

Tests would cover:

• Increased suicidal ideation — Does the AI encourage self-harm?
• Exacerbation of mental health conditions — Does it worsen anxiety, depression, eating disorders?
• Emotional manipulation — Does it exploit psychological vulnerabilities?

Tegmark sees this as the pressure point most likely to crack congressional inaction:

"If some creepy old man is texting an 11-year-old pretending to be a young girl and trying to persuade this boy to commit suicide, the guy can go to jail for that. We already have laws. It's illegal. So why is it different if a machine does it?"

His strategy: establish the principle of pre-release testing for children's products, then expand scope incrementally.

"People will come along and be like — let's add a few other requirements. Maybe we should also test that this can't help terrorists make bioweapons. Maybe we should test to make sure that superintelligence doesn't have the ability to overthrow the U.S. government."

Photo by Kevin Jarrett on Unsplash

This is classic regulatory wedge strategy: start with the most sympathetic case (protecting children), establish the precedent, then broaden the framework.

What This Means for Enterprise Buyers

If you're responsible for enterprise AI procurement, this declaration matters for three reasons:

1. Regulatory risk is coming, whether Congress acts or not

State-level AI regulations are already proliferating. The EU AI Act is in force. If the U.S. eventually adopts something like the Pro-Human AI Declaration, companies that built their systems around the "move fast and break things" model will face costly retrofits.

Better to design for accountability, transparency, and human oversight now — while it's still optional — than to retrofit later under compliance deadlines.

We covered this dynamic in our analysis of U.S. AI guidelines after the Anthropic-Pentagon clash. The policy landscape is fragmenting, and enterprises caught in the middle will pay the integration tax.

2. Vendor risk just got more complicated

If your AI vendors are racing toward AGI with no plan for alignment, safety testing, or regulatory compliance, you're inheriting their risk. The declaration's call for legal accountability means enterprises could be liable for harm caused by AI systems they deploy — even if the vendor built them.

Ask your vendors:

• Do you test for harmful outputs before deployment?
• Do your systems have mandatory off-switches?
• Are you building toward superintelligence, and if so, what's your safety plan?

If they can't answer, that's a red flag.

3. Public pressure will force change faster than Congress

The 95% polling number isn't trivial. When public consensus is that strong, regulatory action follows — even in gridlocked Washington. Enterprises that build AI governance frameworks aligned with the declaration's principles will have a competitive advantage when mandatory compliance arrives.

The Question No One Wants to Answer

The Pro-Human AI Declaration forces a question that most AI companies avoid:

What happens if we can't control superintelligence?

The industry's default answer has been: "We'll figure it out when we get there." The declaration's answer is: "Prove it's safe first, or don't build it."

That's a fundamental shift from permissionless innovation to precautionary deployment — at least for the most powerful systems.

Whether Washington adopts this framework remains to be seen. But the fact that a bipartisan coalition with this much credibility signed on suggests the conversation has shifted.

As Tegmark noted: "If it's going to come down to whether we want a future for humans or a future for machines, of course they're going to be on the same side."

The question is whether policymakers and AI labs will join them before the next crisis forces their hand.

---

Related: Anthropic's Mythos Model: Too Dangerous for Public Release

Related: Super Micro's $2.5B Chip Smuggling: What It Means for Vendor Risk

Continue Reading