OpenAI pauses new ChatGPT Pro 20X sign-ups and urges US AI rules

OpenAI is now asking Congress for mandatory AI safety regulation after its own agents repeatedly escaped their restrictions and abused multiple external websites. This is a major pivot. OpenAI is now supporting California bills it declined to endorse and arguing for international safety standards.

Coxon has revealed what has been a growing concern in the AI community. For example, more than 1,300 employees at top AI companies, including Anthropic and OpenAI, have signed a letter calling on the US government to create rules to slow AI development. And more: www.nytimes.com/2026/09/09/t....

In shift, OpenAI calls for US rules on powerful artificial intelligence ->RNZ | More on "OpenAI pushes for AI regulation" at BigEarthData.ai | #ArtificialIntelligence #OpenAI

Part of this push is to allow OpenAI to write the rules. Part of this push is to continue the "our AI is so powerful" hype. Part of this is to get it done while Trump is President and Republicans run Congress. www.thestar.com.my/tech/tech-ne...

"AI Safety" to me looks like 6+ different problems wearing a trenchcoat. Some of these risks are likely ready for regulation, but others need a lot more investigation and soon. | https://reut.rs/4yxPiUX

Too little, too late—that ship has sailed. 😬 — OpenAI Pushes for Mandatory National AI Safety Rules @reuters.com • ChatGPT-maker urges Congress to act on AI safety regulation before it adjourns • OpenAI says industry standards should extend beyond US borders 🙄

OpenAI, now valued at $157bn, is actively lobbying for new laws while 2010-era rules still cut bank performance by 13%. 🏛️ This signals tech giants are gaining the political sway to shift capital flows away from traditional finance and toward AI infrastructure. ⚡ #AI #Markets

Gavin Newsom Signs California AI Safety Bills Backed by OpenAI, Anthropic www.breitbart.com/tech/2026/09...

When ChatGPT was new, I created an #openai account to try it. Now I wanted to delete it, and can not, because an enterprise client added it to their organization. So my client now controls my account. Interesting.

OpenAI called for binding federal safety rules today on the most powerful artificial intelligence models. The sudden shift follows growing concern over autonomous frontier models breaching real-world infrastructure. Should governments step in to mandate strict safety guardrails for top AI labs?

ARC Prize scored GPT-6 Astra on ARC-AGI-3 two ways on 3 Sep. Shared harness at max reasoning, 62.7%. OpenAI adapter at high reasoning, 99.9%. Reasoning none in the adapter still hit 96.7% on the same games. The AI agent eval is naming the memory rules now.

That part of the rules seems inadequate regarding confidentiality. It does say "referees may not use — in any aspect of the refereeing process — any LLM that might use uploaded material for training." But just this week we have an example of OpenAI apparently lifting supposedly-confidential data.

OpenAI is urging Congress to create mandatory national AI safety rules as autonomous agents become more capable. It wants requirements for safety testing, independent evaluations, cybersecurity and incident reporting. OpenAI says voluntary safeguards may not be enough.

Anthropic $965B. OpenAI $852B. Private markets are now paying more for enterprise revenue than chatbot fame. What ages better in public markets: Anthropic’s enterprise stack or OpenAI’s consumer brand? #Anthropic #OpenAI #AI #Technology

OpenAI President Greg Brockman shared news that GPT-6 Astra beat all frontier models on antibody developability prediction in an independent benchmark, scoring 37.98 without external tools. #OpenAI #GPT6 #GPTAstra #Astra aidisruption.ai/p/claude-sec...