15 AI stories explainx.ai reported on November 12, 2026 so far, ranked by reader interest and grouped by topic. Each links to the full write-up with sources.
Commerce is reportedly finalizing a rule to block Chinese AI firms from remotely using advanced Nvidia chips in foreign data centers; it is unpublished, and only anonymous sources confirm it.
ForeSpark, code from the paper arXiv 2611.13442, sets draft length and verification budget together per request on vLLM and reports up to 1.77x the throughput of DSpark at matched per-user speed.
The Intercept reports that Russian prank duo Vovan and Lexus posed as a Ukrainian intelligence chief and got three OpenAI national security leaders to discuss military AI for Ukraine; OpenAI confirms it was a prank.
AWS added a standalone InvokeWebSearch API to Bedrock AgentCore: one SigV4-signed call returns up to 25 ranked results from an AWS-operated index, at 10 requests per second by default.
Reuters sources say Chinese AI firms are pre-booking overseas Nvidia cloud capacity while fearing new US limits on remote GPU access tied to the Huang scandal and RASA.
OpenAI signed a child-safety charter at the Paris Peace Forum and pledged $2 million to shared measurement tools; the coalition wants pre-market testing of child-facing AI, but no binding rules exist yet.
Financial News quotes Altman warning Samsung that AI memory and data-center demand may be double-counted in the neocloud stack, while he still expects long-term token demand to explode.
Foxconn earned about NT$77.5B in Q3 2026, up 34%, on revenue up 47%; AI racks drive it, but its yearly outlook was repeated, not clearly raised.
Nvidia says Jensen Huang resigned immediately on November 12, 2026; Colette Kress is interim CEO and the board is searching for a permanent successor.
Jim Himes said the Gang of Eight briefing did not settle whether Alice Huang knowingly spied or was a useful idiot, and that Congress still lacks answers on influence, information, and FBI follow-through.
ElevenLabs Scribe v3 and v3 Live improve accuracy and speaker labels, and v3 Live uses conversation context so voice agents wait through mid-sentence pauses.
TensorFold is a Zig-native LLM server for Metal and CUDA that pulls Hugging Face checkpoints and exposes OpenAI-compatible APIs without a Python serving stack.
Hugging Face PyBot is a weight-free Python chatbot agents wrote to 26% on SmolTalkBench, showing how far rule systems go before you need a real model.
Omnii Defense is Radical Numerics' offline screening layer on Omnii-1; Wired confirms the launch and threat-score framing, but independent validation of bioweapon claims is still pending.
Bloomberg reports Pony Ma said Tencent would redeploy 50% of staff to AI-native businesses; the words come from unnamed attendees, a Tencent source denied layoffs, and no plan or timeline is public.
AWS published a guide for running OpenAI Codex against Amazon Bedrock models using AWS single sign-on, with four rollout paths and a reference repo; pricing and quotas are not in the post.
+9 more updated posts
Get each day's AI news in your feed reader: daily RSS · every post