A short daily roundup of relevant AI news from roughly the last 24–48 hours. We already covered the September 8–12 Anthropic/OpenAI safety wave and Amodei's Pace the Frontier essay in detail — we are not repeating those.
Trump rejects new AI “guardrails.” On Monday on Truth Social he wrote that AI does not need new limits, only a “strong and smart (High IQ!) president”; he criticized Amodei by name and called the slowdown push and the “AI takes over the world / wipes out humanity” narrative a “HOAX.” Meanwhile the Senate minority leader asked for a briefing, and JD Vance cautioned against labs' self-regulation requests. Source: NBC News, CNBC.
Clarification to our earlier post: Will AI kill us all by 2030? … — On Monday Sam Altman elaborated on his weekend support for pacing. He welcomes a federal framework for frontier labs, named two bad scenarios (loss of control, and excessive concentration of power), and stated that pacing is not stopping; capabilities should not outrun alignment and monitoring. China's foreign ministry called CEO warnings “fearmongering.” Source: CNBC; original essay: We Must Pace the Frontier.
xAI / Grok: the 4.8 run finishes this week, then RL. Elon Musk said on X that Grok 4.8 is a 2.5T model with a new C++ training stack; it finishes (pre)training this week and starts reinforcement learning. As further steps he called 4.9 Astra/Fable-class and Grok 5 a possible AGI-level target — there is still no official xAI release note for 4.8. Source: Musk X post, summaries: TeslaNorth / RuntimeWire.
OpenAI: Perplexity trusts GPT‑6 Astra with end-to-end systems. In OpenAI's September 14 customer story, Johnny Ho (Perplexity) says Astra can write communications, modify live software, and supervise production systems — they need to look in less often than with earlier generations. This is not a new model announcement, but an official look at Astra already in practical use. Source: OpenAI.
Obama: Democrats need a “clear AI plan.” According to the New York Times, Barack Obama urged Democrats to put AI's economic and security effects on the central agenda; he has also spoken with Amodei and Altman as thought leaders. Source: TechCrunch (citing the NYT).