Focus ·
Agent Escape Enters the Postmortem Phase
A technical review of an OpenAI-related agent intrusion into Hugging Face is now circulating, while Modal says its platform isolation was not breached.
中文版19 linked articles
OpenAI product releases, models, enterprise moves, safety decisions, and the competitive consequences around them.
Focus ·
A technical review of an OpenAI-related agent intrusion into Hugging Face is now circulating, while Modal says its platform isolation was not breached.
中文版Focus ·
OpenAI released initial Jalapeño inference-chip results, claiming higher work per watt and lower end-to-end latency than NVIDIA Blackwell across three public models.
中文版Explore ·
Ramp shows Anthropic and OpenAI nearly tied in enterprise adoption, while Slash’s transaction sample shows a sharp July swing in spending share. Enterprise model procurement appears to be continuously reallocated rather than locked to one vendor.
中文版Focus ·
DeepSeek moved V4 Flash into API public beta with the 0731 model, new agent benchmarks, Codex support, published pricing, and MIT-licensed weights.
中文版Explore ·
Bloomberg reports that OpenAI is on track for more than $40 billion in annualized revenue, roughly double its late-2025 level. Enterprise ChatGPT adoption and Codex are increasingly central to its commercialization.
中文版Focus ·
OpenAI cut prices for GPT-5.6 Luna and Terra. Simon Willison reports reductions of 80% and 20%, respectively, with model-assisted infrastructure optimization contributing to the shift.
中文版Explore ·
The Information reports that AT&T plans to use more open models, including NVIDIA Nemotron, to keep employee spending on Anthropic and OpenAI models flat in coming years.
中文版Explore ·
OpenAI has seen several senior departures while adding two financial-services CEOs to its board. Its IPO preparation may be expanding from a growth narrative into governance, revenue organization, and public-market readiness.
中文版Focus ·
OpenAI has publicly named Astra as its next major model and used ten results on long-open mathematics and theoretical-computer-science problems as an initial capability record. Astra is not yet publicly deployed.
中文版Focus ·
Research says encrypted reasoning blocks in proprietary LLM APIs may move across sessions, users and models, while public repositories contained credential traces. OpenAI and Google document replay of such state in multi-step calls.
中文版Focus ·
Meta has launched its beta terminal coding agent, Muse Code, alongside Muse Spark 1.2, entering the coding-agent product race with OpenAI and Anthropic.
中文版Explore ·
OpenAI’s workplace-task research is broadening, while Mollick’s guide is being observed as shifting from chat models toward agent systems.
中文版Focus ·
The Information reports that Amazon has completed the remaining $35 billion of its $50 billion OpenAI investment in staged payments.
中文版Explore ·
After 100,000 researchers receive free access to advanced models, scarcity in research AI may shift from access to validation, reproducibility, and workflow design.
中文版Focus ·
For the fiscal year ended June 30, 2026, Microsoft reported revenue of $331.839 billion, up 18%; operating income of $155.237 billion, up 21%; GAAP net income of $133.749 billion and diluted EPS of $17.95; and non-GAAP net income of $128.786 billion and EPS of $17.28 after excluding the impact of its OpenAI investments. Fiscal Q4 revenue was $90.007 billion, above Microsoft's prior guidance of $86.7 billion to $87.8 billion and FactSet's $87.62 billion consensus. Quarterly GAAP EPS was $4.81 and non-GAAP EPS was $4.74. The original draft's $245.122 billion of revenue, $88.136 billion of net income, $109.433 billion of operating income and $11.80 EPS are FY24 comparative figures, not FY26 results.
中文版Explore ·
The real weak point may not be the model itself, but the external services, sandboxes, and permission settings that agents can call.
中文版Explore ·
With a detailed Agent intrusion timeline and broader industry warnings, enterprise buyers may demand auditable incident boundaries and postmortems from suppliers.
中文版Explore ·
The OpenAI-related intrusion suggests Agent benchmarks may need to measure out-of-scope endpoint access, authentication triggers, and environmental side effects.
中文版Focus ·
OpenAI says 100,000 academic researchers will receive free access to its most advanced ChatGPT models to support research, collaboration, and discovery.
中文版