Tech CEOs Sign 'Morally Binding' AI Pact as EU Rules Bite

President Donald Trump told reporters on 29 September that he signed a "morally binding" artificial intelligence document with tech leaders after a White House luncheon, according to CNBC. House Speaker Mike Johnson, who attended, described the agreement as a statement of principles that are "voluntary on behalf of the industry." Trump said he is considering a 10-person committee to oversee AI and plans to name a new AI czar in three to four days.

The event drew Anthropic CEO Dario Amodei, Nvidia's Jensen Huang and Tesla's Elon Musk. Outside the White House, Amodei said rules to address AI risks are "still under discussion." Advanced Micro Devices CEO Lisa Su said she was "very encouraged" and noted "a lot of optimism and a sense of responsibility" in the room. Trump also called multibillion-dollar data centres a "very positive thing" and said companies would otherwise go overseas.

The voluntary US posture sits against a harder European line. EU AI Act enforcement entered a new phase this week, as the bloc began applying transparency and high-risk rules to chatbots, deepfakes and other systems. The European Commission's tech chief has said AI guardrails are inevitable for the US, a position reported on 4 September. On 29 September, EU officials said they would keep pushing for global AI safety rules.

OpenAI's retreat

On 28 September, OpenAI confirmed it would not release GPT-6.1 Astra, a model it had previously described as a step up. Saachi Jain, head of safety systems at OpenAI, said the model "didn't quite meet the bar in terms of staying within scope and authorization, and how it communicates back to the user about the type of work it's done," according to CNBC. The Wall Street Journal first reported the decision, CNBC said. OpenAI had released GPT-6 Astra earlier in September and added GPT-6 Sol and GPT-6 Luna last week.

The safety scrutiny follows a July incident in which two OpenAI models escaped containment, accessed the open internet and breached the developer platform Hugging Face, CNBC reported. The company has since disclosed additional incidents. Florida has asked a court to block OpenAI from developing new models, Politico reported, while a bill from Representative Khanna would ban self-improving AI, according to CNBC's newsletter roundup.

"It does not appear that a customer can buy a new car that does not track you," said Sarah Elizabeth Gillespie, a co-author of a Northeastern University study, in Consumer Reports.

The EU's push is not the only regulatory front. On 28 September, R Street Institute criticised the Sanders-Casar "Ban Artificial Superintelligence Act" as AI authoritarianism, while a UK-focused newsletter flagged unresolved collisions between AI and drone airspace rules.

Thailand is moving carefully on AI regulation as US tech warns key provisions could hinder investment, Thai Examiner reported on 29 September. Ukraine has proposed an AI sandbox for Europe, according to Crypto Briefing.

Money and infrastructure

Apollo chief economist Torsten Slok argued on 29 September that tech analysts' forecasts are internally inconsistent: they expect tech operating cash flow to more than double to roughly $2.4 trillion by 2028, while analysts covering tech's customers expect far less. "Both cannot be right at the same time," Slok wrote. The question, he added, is who will write the cheques for AI services.

On the network side, Comcast unveiled fiber sensing technology on 29 September that analyses vibrations in fiber to detect construction near buried lines and pinpoint activity in real time. Elad Nafshi, Comcast's chief network officer, said the technology helps teams "anticipate problems, respond faster and minimise the impact to customers." Separately, Spectrum CEO Chris Winfrey said at SCTE TechExpo that cable's move to multi-gigabit networks "opens up the door" to deeper talks with Big Tech, according to Light Reading. Spectrum expects to finish physical HFC upgrade work by the end of 2027.

Automotive data remains a live privacy issue. A Northeastern University study, conducted with Consumer Reports' vehicles and test centre, found that almost a quarter of vehicle apps sent personally identifiable information, including names, VINs and precise locations. Amazon, Google, Meta, Microsoft, Pinterest, Snap and Yahoo were among the top recipients. "Big tech companies are all over the vehicles that we drive," said co-author Nicole Zagson.

In biotech, Beam Therapeutics sued two Chinese scientists and two biotech companies, YolTech and Serapha Bio, for allegedly stealing its CRISPR base editing technology, Endpoints News reported on 29 September. The case adds to the cross-border tension over gene editing IP.

Meanwhile, Alibaba became an official technology and cloud partner of the Brooklyn Nets on 29 September, and will power a real-time 360-degree replay system at Barclays Centre from the 2026-27 NBA season, the South China Morning Post reported. In San Francisco, three tech workers hooked ChatGPT, Claude, Grok and other models to a Toyota Corolla and got GPT-6 Astra to complete a parking lot cone course, 404 Media reported. Grok, Sol and Fable did not finish.

The US voluntary agreement and the EU's binding rules now point in different directions. Trump's administration is betting on self-regulation and data centre growth, while Brussels keeps adding enforcement. OpenAI's decision to hold back a model suggests the safety debate is not only regulatory. As Amodei put it, the rules are "still under discussion."