4:21 PM
Today in brief.
The AI 2027 team just published the sequel: AI 2040
Daniel Kokotajlo and the AI 2027 team - the group whose forecasting blog has been eerily accurate so far - released "AI 2040: Plan A," a proposal for how international superintelligence development could actually go right.
Meta's coding model wants to compete with GPT-5.5 and Opus
Meta released Muse Spark 1.1, an agentic coding model on its new Model API, and early framing has it competitive with GPT-5.5 and Claude Opus 4.8 - and it's scoring SOTA on HLE.
Grok 4.5 landed and it's aimed straight at coding and agents
SpaceXAI's Grok 4.5 launched today, trained on NVIDIA's newest GB300 NVL72 systems and purpose-built for coding and agentic work - and early testers are already pointing it at real dev tasks with Cursor.
OpenAI shipped the GPT-5.6 family, and it's already the default everywhere
OpenAI launched Sol, Terra, and Luna (GPT-5.6) plus a new autonomous agent called ChatGPT Work and a voice model called GPT-Live - and within a day it's Microsoft Copilot 365's 'preferred model' and Perplexity's default.
+59 sourcesOpenAI's week of turbulence: a departure, a browser shutdown, and a lawsuit going sideways
Same week as the GPT-5.6 launch, OpenAI is dealing with its No. 2 stepping down for health reasons, quietly killing its AI browser, and a New York Times report claiming it hid evidence in the ChatGPT copyright trial.
AI is now a named reason in 31% of June's layoff announcements
Microsoft is cutting 4,800 jobs as it restructures around AI, and separately, AI was explicitly cited in 14,029 layoff announcements in June alone - about a third of the month's total.
Two fresh AI raises: a $1bn coding-agent bet and a Paris voice startup backed by Nvidia
Prime Intellect hit a $1bn valuation on a $130m Series A, and Paris-based voice-AI startup Gradium added $100M in seed funding with Nvidia among the backers.
No stories in this category for this edition.