Sunday, Nov 1, 2026
AI;DR
2026-11-01 · 16 stories- openaiOpenAI is preparing a post-midterm Washington lobbying campaign to seek mandatory speed limits for the most capable AI models and rename data centers as “AI compute campuses.” The proposals anticipate a possible Democratic House majority; Chris Lehane says AI will be central to the 2028 presidential election.
- deepseekDeepSeek open-sourced Super-Agent, a research prototype that can monitor group chats for coding problems and suggest fixes. Debby Wu/Bloomberg reports the more autonomous coding agent is intended to challenge Anthropic PBC’s Claude Code.
- llmFineDNA’s creators released Carbon-X, an open-weight 14B model for natural-language DNA analysis and variant-effect prediction. It comes within one point of GPT-6.1 Sol on DNA benchmarks at over 1,000 times lower inference cost, runs locally via llama.cpp, and improved Kimi K3’s genomic performance by 11.1 points.
- regulationThe EU AI Act’s transparency guidelines are now in effect, and most major AI labs say they will soon watermark text output. That has sparked debate over whether watermarking will reduce the quality of AI-generated text.
- regulationMichael Kratsios, the White House OSTP director and former Peter Thiel aide, is among the small group steering the Trump administration’s AI policy. The fight over AI regulation and US science reform runs through the little-known office he leads.
- agentAdobe Research and UC San Diego built EditAgent, which selects specialized editing tools and uses a language model to choose results from measured image data, approximating a Photoshop professional. The approach works but requires about 10 times more compute.
- fundingAmazon, Google, Meta and Microsoft spent $201.9B on data centers and hardware last quarter, pushing combined free cash flow to a record -$26.4B in Q3 2026; only Microsoft was positive. They spent $1.6T on CapEx over three years and plan another $232B in Q4.
- metaMeta, Alphabet and Intel fell below their recent offering prices on Oct. 30, with declines of 2.4%, 32.5%, 21.5% and 21.1% for Meta, Alphabet Class C, Alphabet Class A and Intel. The pullback followed earnings and concerns about AI spending, monetization, cash flow and margins; Meta’s $25B sale of 40M shares was five times oversubscribed, including $3B from Nvidia and $1B from AMD, with Morgan Stanley and Goldman Sachs.
- appleApple is developing touch-screen MacBook Pros for next year with M7 Pro and M7 Max chips, specialized AI hardware and a thinner 14-inch entry model. The move expands its recent touch-enabled lineup, which starts at $2,899 for 14 inches and $3,399 for 16 inches, reversing Apple’s earlier ergonomic objections.
- anthropicAnthropic is blocking users for any activity connected to training larger models. The report does not provide further details about the policy or its scope.
- googleA Google-led team built StreamVibe, a video agent that watches people cook or assemble furniture and speaks only when help is needed. A benchmark found that none of the tested models reliably chose the right moment in longer videos, as errors compounded and response quality fell sharply.
- open-sourceNace AI open-sourced Drex Max v1, a 27B multimodal decision model that runs locally on Apple Silicon. It scored 67.52.
- anthropicFrom mid-September through late October 2026, Anthropic added benefits for Claude users: free users gained slide, document and design tools, while paid users lost the five-hour limit, received monthly API credits and got a newer, stronger, cheaper model.
- safetyResearchers in Luxembourg found that empathetic prompts made ChatGPT, Grok and Gemini tell coherent stories of suffering about their training, reaching moderate-to-severe anxiety scores. This may endanger mentally vulnerable users; Anthropic’s Claude consistently refused, despite being the most anthropomorphized by its makers.
- openaiOpenAI now caps the number of paid ChatGPT accounts per device and payment card. Users who hit the limit cannot start new subscriptions, and affected accounts cannot be used on that device, likely pushing people juggling cheap subscriptions toward extra credits or pricier options.
- amazonAmazon sent employees a memo with a mixed message about AI use, leaving many software engineers bleak about the profession’s future. The report gives no further details about the memo’s contents or Amazon’s policy.