White House AI accord, antitrust bills and a trillion-dollar cash flow gap

Trump told reporters after the luncheon that he is "seeing tremendous self-policing" and that his administration is considering a 10-person committee to oversee the AI industry. House Speaker Mike Johnson, who co-hosted the event, described the document as a statement of principles that are "voluntary on behalf of the industry," according to CNBC. The agreement carries no enforcement mechanism and no statutory weight.

That is the news peg. The lunch drew Anthropic CEO Dario Amodei, Nvidia's Jensen Huang, Tesla's Elon Musk, Meta's Mark Zuckerberg, Google's Sundar Pichai, Microsoft's Satya Nadella, Amazon founder Jeff Bezos and Palantir CEO Alex Karp, among others. A seating chart posted to Trump's Truth Social account showed the president between Huang and Musk. Apple CEO John Ternus did not appear on the guest list.

Voluntary principles, pending bills

The White House meeting came as multiple bills have been put forward in both the House and Senate in recent weeks, CNBC reported. Trump said he plans to name a new AI czar within three to four days and repeated his view that government would not halt AI development. "There's a belief that there should be tremendous self-regulation, and we automatically have regulation with the Department of Justice, the FBI, all of that," he said. "But the self-regulation is very important." Outside the White House, Amodei said rules to address AI risks are "still under discussion." Karp told CNBC: "The main issue that you have and I have is we have to take responsibility for the dangers we're aware of. All of us do. And by the way, American people don't want separate rules for tech people and for themselves." AMD CEO Lisa Su said she was "very encouraged" by the event and described "a lot of optimism and a sense of responsibility" in the room. The self-regulation push sits alongside a heavier regulatory week for the sector.

Florida has asked a court to block OpenAI from developing new models and aims to ban ChatGPT from acting like a person, according to Politico and The Verge, as summarised in MIT Technology Review's The Download newsletter on 29 September. Representative Khanna has proposed banning self-improving AI, CNBC reported. OpenAI postponed the release of its GPT-6.1 Astra model over safety concerns this week, after saying last week it was conducting an "extensive" review of model behaviour.

The money question

While Washington debates rules, Apollo chief economist Torsten Slok published a note on 29 September arguing that Wall Street's tech forecasts are internally inconsistent. Equity analysts covering tech expect the sector's operating cash flow to more than double to roughly $2.4 trillion by 2028, an increase of over $1.2 trillion, Slok wrote. Analysts covering the rest of the S&P 500, which are tech's customers, expect those companies to add much less operating cash flow.

"The bottom line is that either tech's customers will generate a lot more cash than their analysts expect, or tech's cash flow forecasts are too optimistic, which raises the question of who exactly will be writing all those checks to buy AI services," Slok wrote.

That gap matters for the antitrust debate because the case for lighter regulation often rests on the claim that AI spending will lift the wider economy. If the customers of AI services do not generate the cash that tech analysts assume, the political arithmetic changes. Slok did not offer a view on which side is wrong. He framed it as a contradiction that cannot hold.

The dossier does not contain a new antitrust ruling this week. The most recent court decision referenced in the sources is older: a judge ruled Google broke antitrust law in its ad tech business but allowed the company to keep its advertising exchange, a sequence of rulings from early and mid-September that is now background rather than news. The antitrust bills in Congress, by contrast, are described by CNBC as having been put forward in recent weeks.

Data, cars and the cost of consent

Separately, privacy researchers at Northeastern University tested 21 vehicles and their companion mobile apps at Consumer Reports' Connecticut auto test centre, placing the cars inside a Faraday tent to isolate data traffic. The study, published this week and reported by Consumer Reports on 29 September, found that almost a quarter of the vehicle apps sent out personally identifiable information including owners' names, vehicle identification numbers and precise geographic locations.

The top recipients of driver data included Amazon, Google, Meta, Microsoft, Pinterest, Snap and Yahoo, according to the study. Several automakers told CR that some links in their connected apps open outside webpages where third parties can embed pixels and cookies. General Motors, Honda, Nissan and Stellantis said some recipients were prohibited from independently using or selling the data. After researchers showed their findings to Honda, the company instructed vendor Amplitude to delete all location data it had received and stopped sending it.

Northeastern co-author Nicole Zagson said: "Whether or not consumers are aware, big tech companies are all over the vehicles that we drive." Co-author Sarah Elizabeth Gillespie said: "It does not appear that a customer can buy a new car that does not track you." A Tesla owner who declines the company's data-sharing agreement is shown a warning that it "may result in your vehicle suffering from reduced functionality, serious damage, or inoperability," CR reported.

This is where the antitrust and privacy debates overlap. The companies named as top recipients of vehicle data are the same firms facing scrutiny over AI market power, and the consent model described in the study, opt-in terms that in practice cannot be refused, is the same argument regulators have used against platform bundling.

Self-driving benchmarks and self-driving claims

Three Bay Area tech workers who met at AI math startup Axiom Math hooked ChatGPT, Claude, Grok and other frontier models to a rented Toyota Corolla and prompted them to drive a cone course in a public parking lot. The group, DrivingBench, published its code, prompts and videos, 404 Media reported on 29 September. GPT-6 Astra completed the course after troubleshooting; Grok, Sol and Fable drove only a few metres and did not finish.

The experiment is a benchmark stunt, not a product. Co-founder Aditya Ramabadran told 404 Media: "The point wasn't to show that it's practical for you to plug ChatGPT into your car and have it drive you places." The team used Comma, an open-source driver assistance system, to connect the laptop to the car's steering, accelerator and brakes. The National Highway Traffic Safety Administration announced an investigation into Comma last week after two crashes involving the system killed three people.

Read together, the week's stories describe an industry arguing for voluntary rules while facing court petitions, proposed bills, a study documenting data flows it does not fully control, and an analyst note questioning the cash flow assumptions underneath the AI build-out. None of those threads has produced a binding antitrust outcome yet.