One Person's Remarks Suddenly Shake a Nation
Jakarta, CNBC Indonesia — The resignation announcement of Jacob Coxon, a researcher at the artificial intelligence (AI) startup Anthropic, has triggered a new wave of concern in the United States. The impact is far from trivial: the US Congress is now moving quickly to discuss legislation requiring AI companies to prove their products are safe from harmful risks.
Previously, Coxon caused a stir after revealing the reason behind his decision to leave Anthropic. He said developers in the industry quietly believe AI could annihilate humankind before the end of this decade.
In response to the escalating concerns, US senators immediately moved to design preventive regulations for high-risk AI models.
According to a Reuters report on Tuesday (15/9/2026), Senate staff and negotiating teams said the rules could grant special authority to the US Secretary of Commerce. The minister would be entitled to press AI companies to show valid evidence of safety systems, and even be empowered to send in independent auditors to directly test their safety.
Nevertheless, deliberations on the regulation remain at an early stage. Reuters noted the bill’s chances of being passed are slim, even if it gets the green light from Congress.
Resistance has also come from the White House. President Donald Trump has stated that existing legal instruments are already sufficient to regulate and take action against the technology giants. The signal shows his reluctance to sign off on new AI legislation.
In addition, negotiators are reportedly discussing involving federal courts to block the release of certain AI models deemed dangerous. On the other hand, companies would retain the legal right to challenge such decisions in court. The details of this cross-agency authority are still being worked out.
Another crucial point in the draft, according to lobbyists and legislative staff, is a clause barring state-level governments from creating their own legal rules to oversee AI risks, in order to maintain uniform national regulation.
Tech giants’ bosses call for AI to be reined in
Not long after Coxon’s admission, the bosses of the technology giants finally also agreed to slow the enhancement of AI. Anthropic CEO Dario Amodei published a long essay posted on the platform X on Saturday (12/9). He openly called on the technology industry to stop racing to accelerate AI superintelligence capacity and instead begin applying the brakes.
“We must slow the pace of AI capability improvement. Progress will still look fast, and we must use the time we gain wisely,” Amodei wrote, echoing the stark warning previously issued by his own former researcher, Jacob Coxon, about the potential existential threat to humankind.
To avert potential catastrophe, Amodei put forward a blueprint containing three tactical steps:
Implementation of independent evaluators: placing outside supervisors with full employee-level access to test and verify the safety aspects of systems.
Alliance of AI developers: building close coordination among leading frontier startups to set standardised safety benchmarks while limiting uncontrolled wild experimentation.
Global diplomacy: encouraging cross-country cooperation to formulate a framework for mitigating AI risks at scale.
Amodei’s idea has been endorsed by other global technology giants’ bosses who are also developing AI. Here is a summary:
Elon Musk (xAI): Responded coolly but firmly with a brief reply on X: “Dario is right.”
Sam Altman (OpenAI): Took a more moderate but serious position. He stressed that the concept of “braking” (pacing) does not mean “stopping altogether”, but rather adjusting the rhythm so that safety does not lag behind business ambitions. “The world deserves the assurance that American companies are acting responsibly,” said Altman.
Satya Nadella (Microsoft): Chose an ethical and inclusive perspective. According to him, any pursuit of superintelligence would lose its meaning if it ends up threatening human control. Microsoft is also preparing to release a “Code of Ethics” for its own AI models to open a public consultation process. “The key is that this must not be controlled by just a handful of entities,” Nadella asserted.
When the most influential technology architects on Earth suddenly agree that their own “speed” is starting to feel frightening, the question now is: will this emergency brake truly be pressed, or is it merely a sweet gimmick on social media timelines?
In your view, will this delay in AI development actually be implemented, or will the industry keep charging full speed ahead for business glory?