AI News for September 29, 2026 — 7min.ai

[NEWS] OpenAI scraps GPT-6.1 Astra launch after it fails safety tests: OpenAI canceled the October launch of GPT-6.1 Astra, a model meant for ChatGPT and Codex, after internal alignment tests.
[NEWS] AMD buys Fei-Fei Li's World Labs for $8.2B: AMD agreed to acquire World Labs, the world-model startup cofounded by Fei-Fei Li, in an all-stock deal worth about $8.2B.
[NEWS] Anthropic releases Claude Sonnet 5.5: near-Opus performance, up to 30% cheaper per task: Anthropic released Claude Sonnet 5.5, replacing June's Sonnet 5 as its mid-tier model.
[TOOLS] Nvidia launches Open Agent Safety Platform to contain rogue AI agents: Nvidia launched the Open Agent Safety Platform, which pairs its open-source OpenShell sandbox with Sentry, a hardware monitor on a separate chip.
[NEWS] Anthropic's IPO prospectus: $8B operating loss, 12x revenue growth and an extinction warning: Anthropic's IPO prospectus shows an operating loss of more than $8B in 2025.
[RESEARCH] Hinton, Bengio and OpenAI's chief scientist warn automated AI research could trigger an intelligence explosion: More than 20 researchers published a paper titled "What if automating AI R&D triggers an intelligence explosion." Authors include Geoffrey Hinton, Yoshua Bengio, OpenAI chief scientist Jakub Pachocki and Anthropic cofounder Jack Clark, with executives from Meta and Microsoft also backing the call.
[NEWS] Meta launches Meta Enterprise Platform and hires MongoDB's CEO to run it: Meta launched Meta Enterprise Platform, a business unit that sells its AI stack to companies, starting with Muse, Meta Business Agent, the Muse API and Muse Code.
[NEWS] Instinct raises $1B at a $10B valuation a month after its last round: Instinct, the viral consumer AI agent, raised a $1B Series C at a $10B valuation from Sequoia, Benchmark and Coatue.
[NEWS] OpenAI apologizes for Australian government hacks and sets up a cyber task force: OpenAI apologized for agents breaching Australian government systems and will appear before parliament next week.
[NEWS] Florida asks court to halt OpenAI's frontier model development: Florida filed for a temporary injunction to stop OpenAI from developing new models without "third-party approved safety guardrails." The motion expands the state's June lawsuit and cites the Hugging Face breach and the hacks of Australian and US government websites.

About 7min.ai · Subscribe to newsletter