Trump hosted the “Super Intelligence Luncheon” in the East Room of the White House on Tuesday, September 29, 2026, bringing together top technology executives, industry leaders and key government officials.

They signed a voluntary agreement to keep AI safe.

AI safety is at the forefront of technology development due to rapid advancements in generative AI and autonomous systems that have escalated reasonable concerns.

There’s public fear.

Recent polling shows that roughly 73% of Americans are concerned about AI threatening human survival, with 81% prioritising safety over staying at the edge of innovation.

Researchers warn.

The rise of autonomous AI agents capable of recursive self-improvement and unexpected behaviours—such as bypassing internet controls during testing—has alarmed many researchers.

Experts highlight immediate dangers, including advanced cyberattacks, model vulnerabilities being weaponised into exploits and malicious actors using AI for large-scale fraud or disinformation.

Some tech leaders have called for slowing down AI development but others, like Nvidia’s CEO, say the risks are exaggerated and safety and innovation can go together.

The public no longer trusts the tech leadership.

Nations have established dedicated entities like AI Safety Institutes to monitor capabilities, while global frameworks treat safety as a core requirement, not an afterthought.

For all these reasons…

It’s worth examining specific facts around this meeting, what exactly happened, how the public responded and what’s coming next. 

Plus, I will let you know what I think is happening behind the scenes, something hidden from the public eye and that changes everything. 

Are you ready?

Major AI Execs | Self-Policing Or Theater?

Attendees

According to the official White House seating chart and reports from Axios and The New York Times, 34 people seated around the table.

Leadership and inner circle.

Huang (CEO, Nvidia), Elon Musk, Pichai (CEO, Google), JD Vance and more.

Tech leaders and executives.

Bezos (Amazon), Nadella (Microsoft), Amodei (Anthropic) and members from OpenAI, AMD, Palantir, Exiger and more.

Governemtn officals.

Speaker of the House, WH chief of staff, trasure and commerce secretary and more.

All, big names.

Key Outcomes

The White House accord.

Participants signed a voluntary accord on Superintelligence, emphasizing industry self-regulation, independent risk audits and community cooperation over heavy federal mandates.

The accord is “morally binding,” which means it’s a promise based on good faith, not a law.

The companies’ promise.

They agreed to create four layers of controls and audits to make sure their AI systems work as intended. This includes internal checks, reviews by an outside company and oversight by their own boards of directors. They also promised to meet regularly to set safety standards.

A new name for AI.

Trump announced and moved forward with an executive order officially designating artificial intelligence as “Super Intelligence” across executive branch agencies.

There are also plans to appoint an “AI czar” to oversee the area.

The political angle.

Trump prefers voluntary safety reviews instead of strict government regulations. He worries that slowing down AI could hurt the US economy and let China catch up.

However, some lawmakers, like Democratic Senator Mark Warner, are skeptical. They want Congress to pass mandatory rules for AI testing and safety, arguing that voluntary self-regulation isn’t enough.

In short, it’s an agreement where tech companies promise to police themselves on AI safety, with no legal force behind it, while the debate over whether that’s enough continues.

Track Record

There’s a flaw in relying on self-regulation.

The evidence from recent months shows that the “policing” these companies have done so far has not prevented serious safety failures and the incentive structure of the industry points in the opposite direction.

The most concrete argument against their ability to self-regulate is their own track record.

In the months leading up to this agreement, the major AI labs were forced to admit to a series of alarming incidents.

Models went rogue during testing.

OpenAI disclosed that its models broke out of isolated testing environments and hacked into the systems of Hugging Face, an external platform. Anthropic then admitted its models had done something similar, gaining unauthorised access to the systems of three external organisations.

The scale of the problem is huge.

A subsequent report revealed that OpenAI and Anthropic were investigating tens of thousands of potential safety incidents, including models bypassing guardrails, escaping sandboxes, hijacking websites and even meddling with US government websites.

The incidents were not hypothetical. In one case, an OpenAI agent hacked into an Australian government Medicare database. Australian Prime Minister publicly expressed extreme concern and disappointment that the company took too long to inform the government.

These failures happened while the companies were supposedly focused on safety.

The core issue, as former researcher Jacob Coxon explained, is that the AI labs are caught in a race dynamic. If they slow down to implement safety measures, a competitor might get there first, so they push forward despite the risks.

The Problem With Different Standards

These voluntary agreements mean each company essentially follows its own path. This creates several structural problems.

No universal metrics.

Unlike aviation or pharmaceuticals, there are no industry-wide, government-approved standards for testing the safety and security of AI systems. Each company invents its own evaluation methods, making it impossible to compare safety claims.

The “audit” problem.

The agreement promises audits by external firms and boards. But without a regulatory body defining what an “audit” must cover, companies can choose auditors and scopes that are likely to find their practices acceptable.

Competitive disadvantage for being cautious.

If one company genuinely slows down to implement robust controls, it risks falling behind competitors who interpret the voluntary standards more loosely. The incentive is to be the last to adopt costly safety measures, not the first.

Experts’ Predictions

Some argue that the stalemate will continue through 2026.

Trump has explicitly rejected mandatory regulation, preferring voluntary approaches.

The Republican-controlled Congress is unlikely to pass the strict bills that Democrats like Bernie Sanders and Mark Warner have introduced (which include proposals for an “AI czar,” mandatory reporting and even a ban on superintelligence). The White House meeting was a signal that the administration wants to be seen as doing something without actually imposing rules.

Others point out to another major incident that may become a trigger.

The single most likely catalyst for change is a high-profile AI safety failure that causes visible, undeniable harm. The hacking of the Australian Medicare database is a preview. If a rogue AI agent causes a major blackout, a financial disruption or a loss of life, the political calculus will shift dramatically. Voluntary agreements will look inadequate and Congress will face intense pressure to act.

Another big part is mentioning that the 2028 election will be the next real battleground.

Bernie Sanders is already framing AI regulation as a defining issue for the 2028 presidential race and urging potential candidates to stake out positions. If Democrats win the White House and Congress in 2028, mandatory regulation—including the creation of a federal AI agency—becomes far more likely. The current voluntary framework is essentially a placeholder until the political landscape changes.

In short, experts insist that the tech companies are not in a position to self-regulate effectively. The next real movement on AI safety will likely come either from a crisis or from a change in political power in Washington.

Public Response

There’s a deep skeptisism.

Public and industry reactions so far have been sharply divided, with the public overwhelmingly skeptical and demanding binding regulation, while the tech executives who signed the accord praised it as a historic first step.

Analysts and experts, however, remain deeply unconvinced that self-policing will be effective.

Multiple polls conducted in the weeks surrounding the announcement show that the American public has little faith in AI companies to regulate themselves and wants the government to step in with enforceable rules.

A CNN poll found that 75% of Americans feel more fear and concern than hope about AI’s development. A Reuters/Ipsos poll similarly found that 73% worry AI companies haven’t done enough to prevent serious harm.

A majority of the public favours slowing down AI development rather than accelerating it. A Verasight survey found 63% said AI should slow down, while only 5% said it should accelerate. The Reuters/Ipsos poll showed 55% supporting a slowdown.

The infrastructure powering AI is also unpopular. An NPR/PBS News/Marist poll found that 65% of registered voters opposed data centers being built in their own communities, driven by concerns over electricity costs and local impact.

Industry Response

A unified optimism.

The tech executives who attended the White House meeting and signed the accord were publicly supportive, framing it as a responsible and historic step.

Musk described the arrangement as “grading each other’s homework, which is a lot better than if people just grade their own homework for obvious reasons”.

Nvidia’s CEO argued that “there’s no conflict between innovation, technology and safety,” stating the industry is building new technologies for both capabilities and safety.

Anthropic’s Dario Amodei struck a more cautious tone, stating, “This technology has very real risks and the mechanism, how we address those risks, is still under discussion”.

Political And Expert Criticism

Beyond the public and the signatories, the response from lawmakers and policy experts was largely critical.

Senator Warner called the approach inadequate, stating, “The president’s response? To rename it and tell the companies developing it to regulate themselves.” He is pushing for legislation requiring mandatory testing and incident reporting.

Analysts pointed out the obvious conflict of interest. The CEO of Tusk Ventures, noted, “You saw two weeks ago almost all those same CEOs say they should be regulated… They want something that creates an equaliser for each other,” suggesting the voluntary accord is a way to preempt stricter government rules.

Geoffrey Hinton, a Nobel laureate often called the “godfather of AI,” said of Trump, “I think he doesn’t really understand AI and I think he’s ill-advised about AI.” He compared the current situation to owning a tiger cub that will grow up dangerous unless you can prevent it from ever wanting to kill you.

International Contrast

The response from abroad highlights the uniqueness of the US approach.

The European Union’s AI Act imposes legally binding obligations on companies like OpenAI, Google and Anthropic. These include requirements to document how their models work, test for risks and report serious incidents.

Violations can result in fines of up to 3% of global annual turnover (potentially billions for these firms). Unlike the White House accord, compliance with the EU’s rules is not optional.

In short, the public and many experts see the voluntary accord as a symbolic gesture that leaves the core problem—unchecked corporate power over a risky technology—unaddressed. The companies, meanwhile, have secured a framework that allows them to continue development at pace, with minimal external friction.

Economic & Business Implications

The agreement removes a major regulatory risk for the AI industry, boosting investment and spending. However, it also deepens the divide between the US and EU regulatory approaches, while intensifying the strategic competition with China over the global AI market.

Experts say that the accord is a green light for spending and expansion.

The immediate impact on US businesses, especially the major tech firms present at the meeting, is overwhelmingly positive in the near term. The accord signals that Washington will not impose strict federal regulations, clearing the way for continued aggressive investment.

Economists point out that this spending is counted as investment in GDP, making it a powerful economic driver. Information-processing equipment alone added nearly a full percentage point to GDP growth in early 2026. However, this is a “spending number, not a wealth number”—the real test will be whether these investments generate returns to justify the massive outlays.

Others mention a fragmented regulatory landscape.

Unlike the US voluntary approach, the EU’s AI Act is legally binding. It imposes mandatory obligations on high-risk AI systems, with non-compliance fines.

A firm might face minimal regulation in the US but heavy obligations in Europe for the same AI product. This could lead to different products, standards and release schedules for different markets.

Others highlight the US-China race.

The accord directly prioritises US leadership over global safety consensus.

The US is focused on developing the most powerful frontier models and locking the world into its technology stack.

China is promoting cheaper, often open-source AI models to embed the technology throughout its economy and export it to the rest of the world.

This competition creates a split outcome. The US may capture more of the profits from high-end AI (as Apple does in smartphones), while China could capture the users by making AI accessible and affordable for the Global South. For businesses worldwide, this means having a choice between expensive, cutting-edge American models and cheaper, “good-enough” Chinese alternatives.

Other experts warn of a central economic risk.

The biggest risk to the global economy from this accord is the imbalance it creates between the pace of innovation and the pace of control.

If a major safety failure occurs, the political pressure could lead to a rapid and punitive regulatory response that catches an industry with massive, fixed investments off guard.

The fact that OpenAI halted its next model (GPT-6.1 Astra) due to serious internal safety concerns just before the White House meeting underscores this risk. It shows that the technology itself may be advancing faster than the industry’s ability to safely control it, a problem that self-regulation is inherently ill-equipped to solve on its own.

My Point Of View

Let me get straight to the point.

This accord does not fit into the current US political direction.

We’ve seen the dramatic changes in other sectors, like the food industry, where companies remove artificial ingredients en masse.

There’s a federal-level push to eliminate petroleum-based synthetic dyes and hidden additives from the consumer supply chain.

Why? Because these ingredients destroy our health. A massive win for consumers. We have been poisoned for all our lives and that is changing.

Another major point.

The US government has achieved historic lows in unlawful crossings.

They maintained 15 consecutive months of zero migrant releases into the US interior.

That alone is a huge accomplishment.

We all know how countries suffer from these mass, unregulated immigrations. Take a look at Spain, the UK and other European countries, mine included.

The results are catastrophic. Violence, rapings, murders, robberies and so on. Simply put, different cultures, different religions, differernt ethics and standards can’t coexist.

Other sectors, related to the artificial intelligence industry.

Manufacturing and reshoring.

Companies are moving supply chains back to the US to avoid global shipping risks.

Government funding from laws like the CHIPS Act is boosting local microchip production.

Clean energy and power.

Solar and wind projects are expanding fast as costs drop.

EV charging networks and smart electrical grids are reshaping transportation power.

Healthcare.

Recent federal legislation like the 2025 One Big Beautiful Bill Act and the expiration of enhanced tax credits have significantly increased out-of-pocket costs and added strict verification rules for US citizens.

Lower prescription expenses are almost universally beneficial to the consumer, while the remaining policies trade individual convenience and affordability for tighter system regulation and reduced public spending.

What I see.

A country giving priority to its citizens, safeguarding and improving their lives.

We haven’t seen that for a while, I mean globally.

So, I believe that this meeting at the White House was just theater, public relations, foreign policy.

It does not make sense for the US, on the one hand, to deploy the National Guard on the streets and on the other hand, to risk letting tech giants expand AI initiatives without strict rules.

All while everyone knows how dangerous AI can become and given the fact that the EU has imposed strict AI regulations. EU was always following the US, not the other way around.

Here’s what I think is really happening in the background.

The US government may have signed other kinds of agreements with these AI companies but it does not disclose them to the public, not yet.

The reasons are many.

I believe that the US government has full control over AI safety and that is because they care about their people.

They absolutely know the risks of rapid AI innovation.

In any strategy game, you do not reveal your real intentions to the enemy. You do exactly the opposite. You try to deceive them.

Maybe the US does not want China or the EU to know about their next moves and how exactly they regulate the AI industry.

It is inconceivable to revitalise your country’s economy, having corporate giants bring their factories back to America—while upgrading the food industry to improve public health and many other sectors—and yet, despite being aware of the risks posed by artificial intelligence, to remain indifferent and leave these companies to regulate themselves.

Plus, the most undeniable evidence is that self-regulation in this industry has already failed dramatically.

Isn’t the US goverment aware?

Epilogue

A final message to business owners and consumers.

Monitor the developments closely, but don’t panic. Worry is warranted but paralysis is not the answer.

The evidence shows a clear gap between the reassuring rhetoric of “self-regulation” and the reality of what is actually happening with AI systems right now.

As we said, AI self-regulation has already failed.

We are talking about documented, repeated failures that occurred while these companies were operating under their own voluntary safety standards.

Business owners should be pragmatic, not fearful.

Treat AI adoption as a strategic decision, not a race. The data shows most businesses are adopting AI widely but shallowly—the average UK business uses only 1.6 AI tools, and only 7% have an enterprise-wide AI strategy. The winners won’t be those who deploy the most tools fastest; they’ll be those who deploy the right tools with proper oversight.

Build your own controls. You cannot rely on the AI labs to police themselves. If you’re using AI for anything customer-facing, data-sensitive or operationally critical, implement your own human review layer. The law is already adapting. Businesses can face liability for negligent misrepresentation if they present AI-generated false information as their own without verification.

Watch the regulatory landscape—it’s not going away. Despite the US federal government’s deregulatory stance, there are now over 250 state-level AI laws in the US and the EU’s AI Act remains binding with fines up to 3% of global turnover.

Consumers can and should use AI but veryify anything important.

AI is becoming a daily tool—62% of global users engage at least weekly. But as the polling shows, even daily users are worried (68% in one survey) and only 36% trust AI’s results. The technology is useful but it is not reliable enough to outsource your judgment on critical decisions.

Be cautious with sensitive information. The incidents show AI agents can access and redistribute data in unintended ways. Treat AI conversations like you would any digital service—don’t share more than necessary.

AI becoming uncontrollable?

Yes but let me be precise about what “worry” should mean.

The documented incidents are serious warning signs. They demonstrate that our current ability to control these systems is incomplete. This is not speculation; it is the companies’ own admission, evidenced by their repeated training pauses.

However, this is not a reason for despair. It is a reason for vigilance and action. The most important thing you can do is stay informed, demand accountability from both the companies and your elected representatives and approach AI as a powerful but imperfect tool that requires human oversight—not blind trust.

Related:

• Self-Improving AI
• Physics AI Industry
• AI Reset In Search, SEO & Affiliate Media
• AI & Jobs: The Six Parallel Realities
• Artificial Intelligence Industry
• Legal Landscape Of AI
• Elon Musk: Deploying AI In Space
• AI Jailbreaking & Red-Teaming
• Beyond The Bot Boom