Who sets the rules for AI: Washington renames it, New York subpoenas, DoubleVerify measures ChatGPT ads

The week before, the agent economy gave itself rules, then invoices, then resignations. This Monday the question goes up a level. Who sets the rules of the game for artificial intelligence in public life?

Washington answers with a new name and a new task force. New York answers with a hearing room and subpoenas. The ad industry answers with the first real measurement for buying attention inside the machine.

Signal one: the government renames the technology

On September 29, President Donald Trump signed Executive Order 14434, titled "Inaugurating the Era of Super Intelligence." It directs every federal agency to stop using the term "artificial intelligence" in official documents, websites and reports, and to write "Super Intelligence," or "SI," instead.

The order does not touch the statutory definition. For now, Super Intelligence means exactly what artificial intelligence already meant in US law. The president's science and technology adviser has 60 days to propose a federal definition of the new term.

The task force

On Sunday, October 4, the president added an institution to match the name. On Truth Social, he announced a Super Intelligence Force, a federal task force to coordinate the government's relationships with consumers, public interest groups, religious organizations, critical infrastructure providers and AI companies.

It is led by four officials:

• Jay Clayton, described as Director of National Intelligence;
• Andrew Ferguson, chair of the Federal Trade Commission;
• Emil Michael, Under Secretary of Defense for Research and Engineering;
• Scott Kupor, director of the Office of Personnel Management.

The task force reports to the president and his chief of staff, Susie Wiles. Its stated mission: to make sure America continues to lead the world.

Context matters. A few days earlier, the same companies had signed a voluntary safety accord, which TechCrunch refers to as the Joint Commitment on Frontier Responsibilities, called morally binding by the president and carrying no binding rules. The Seoul Economic Daily's reading is blunt: Washington now treats AI as a strategic asset for national security and competition with China, far more than as an industry to encourage.

The name is already leaking into the private sector

That same Sunday, Elon Musk confirmed on X that SpaceXAI would become SpaceXSI. Reuters independently confirmed the plan. Nothing in the order requires it: it applies to federal agencies, not companies. Musk chose it anyway. His comment: "No more AI. SI, it's better." It is the second name change in under a year, after xAI was folded into SpaceX in February 2026.

What changes

Three things.

Vocabulary is a budget line. The words a government uses decide which department pays, which committee legislates and which courts interpret. Super Intelligence sounds like a space race far more than a tool. The word reframes AI from a product category into a national project. And national projects attract national money and national control.

Look at who is not in the room. The task force is four government officials and no company executives. The pledge phase is ending, the coordination phase is starting, and the state writes the agenda. Whoever sets the standards this winter will write the procurement rules companies sell into next year.

The rename is marketing in service of policy. The name promises a destination before the definition exists. Sixty days to define the word, and a federal brand already built on it. I have worked on names for a long time. I recognize the move: install the perception, then fill in the substance. It often works. It also creates a debt, that of a name promising more than reality will deliver.

For brands, the consequence is practical. When Washington reframes the technology you buy as a strategic asset, your vendor relationships become geopolitical exposure. You need to know where your AI supply chain lives: which models, hosted where, subject to which jurisdictions.

Signal two: New York puts the builders and the whistleblowers in the same room

That day the New York City Council holds its first hearing on AI risk, and the staging is remarkable.

On one side, the companies: Google, OpenAI, Anthropic and Meta. Only Meta agreed to come voluntarily. The others came after Council Speaker Julie Menin sent a letter explicitly threatening to use the council's subpoena power. SpaceXAI did not respond, and the council issued a subpoena to compel it.

On the other side, the whistleblowers. Jacob Coxon, a former Anthropic researcher who resigned the previous month accusing his former employer and OpenAI of "gambling with our lives," testifies at the speaker's request. Alex Turner, a former Google DeepMind researcher, plans to attend. Daniel Kokotajlo, executive director of the AI Futures Project and a former OpenAI researcher, is expected.

The companies send their policy leads: Logan Graham, head of Anthropic's Frontier Red Team, Morgan Dwyer of OpenAI's policy team, Shane Cahill, Meta's AI policy director, and Alice Friend of Google.

The bills on the table

They are concrete. The council is considering legislation that would:

1. create an incentive program for AI whistleblowers;
2. require independent third-party validation of AI systems;
3. study, according to press coverage, emergency kill switches.

All 51 council members are expected. It is the first time the major AI companies have to explain their safety measures publicly before an elected body of this size.

What changes

The pattern repeats, and it is accelerating. California's attorney general demanded OpenAI's incident trail. Congress wrote liability into a bipartisan bill. Now a city council is building the machinery that will make whistleblowers a permanent part of AI oversight. The whistleblower becomes an institution.

I also note the level. A city. AI rules will not only come down from Washington. They will also come up from cities and states, with different requirements. For a company operating across jurisdictions, that means a patchwork of rules to follow, as data protection already showed.

For brands, the lesson from the end of the previous week holds and hardens. Your safety story is the people who write it. Some of them now testify under oath.

Signal three: the first real measurement of ads inside the machine

DoubleVerify, the ad verification company listed on the New York Stock Exchange, announced attribution measurement for advertising in ChatGPT. Advertisers can now treat ChatGPT ads as a touchpoint in the customer journey, measuring their contribution to conversions and return on ad spend through DoubleVerify's Rockerbox platform, alongside every other channel.

The second half of the announcement is the one to watch. DoubleVerify and OpenAI are building a brand suitability evaluation pilot. Independent partners will work with OpenAI to evaluate and help define the policies and guardrails for ChatGPT ads, in a controlled testing environment, without accessing users' private conversations.

Why it matters

Two notions deserve explaining.

Attribution means working out what share of a sale belongs to each advertising touchpoint. Without attribution, a channel is spending on faith. With it, it becomes a line you can manage, compare and grow.

Brand suitability is the set of rules deciding next to what content an ad may appear. In traditional media, it is a list of topics to avoid. In a conversation with an AI, the problem is new: the brand's message appears next to what the model just said, and what the model says changes with every exchange.

Until now, brands bought ChatGPT inventory on trust. An independent measurement layer plus a suitability framework turns a black box into a media channel. This is the moment the attention economy inside AI gets its standards. They are set by a verifier, not the seller, and that matters.

What changes

Money follows measurement. As soon as advertisers can attribute revenue to a ChatGPT touchpoint, budgets can move. And budgets move the product roadmap. The interface people chat in becomes the shelf where brands compete.

Whoever writes the suitability rules writes the editorial policy of the conversation. The pilot is small and controlled. It is still the first version of what brands are allowed to say, and not say, inside the most visited AI product on the planet. I would advise brand strategists to read this pilot's results the way strategists once read Google's first keyword exclusion lists. The guardrails become the brief.

I connect this signal to the first episode of this series. When the agent picks the merchant, a brand has to be readable by the machine. When ads appear inside the conversation, a brand also has to be compatible with what the machine says. Two new requirements that ask for the same thing: a brand clear about what it is and what it refuses.

The thread of the day

Washington renames the technology and installs its task force. New York summons the builders and the whistleblowers to the same room, under oath. The ad industry quietly writes the first rules of commerce inside the machine.

Three different rooms, one negotiation: who sets the rules for AI's presence in public life.

The signal under the noise: the rules are being set now, in vocabulary, in hearing rooms and in measurement contracts. Whoever waits for the final rulebook condemns themselves to reading it instead of writing it.

The noise filtered. The signal kept.