The Brief Brief
the week in AI, briefly. then briefly again.
Artificial Intelligence — briefly, then briefly again · tbb.ceo
listen tothe daily 0:00 –:––
/daily ·06 NOV 2026 ·FRIDAY ·3 MIN READ ·7 STORIES

Markets, models, and an email problem

Asian equities posted their worst AI-linked single-day drop on record, two new frontier models launched, and Claude Code became a privacy story.

01 / The Day

FRIDAY 06 NOV 2026, ranked

07

Asian AI tech stocks suffer record drop on data center fears

Taiwan's Taiex fell a record number of intraday points, South Korea's Kospi dropped 8.7%, and Japan's Nikkei fell 6.7% — driven by investor concern that the Democrats' midterm sweep will delay or block US data center expansion. The sell-off concentrated in GPU foundry suppliers, cloud hardware, and HBM memory producers.

  • Taiex fell a record intraday point drop; Kospi -8.7%, Nikkei -6.7% on data center permit risk
  • Concern centres on Democratic proposals for data centre moratoriums in Texas and federal oversight of AI infrastructure
  • Asian chip and hardware supply chains most exposed — HBM and TSMC-linked stocks led the declines
Why it mattersThe first quantified market reaction to the midterm data centre outcome, reaching hardware supply chains six time zones away.
01 techmeme.com↗

Inkling 2 Small: 594B MoE, 17B active, matches Gemini 3.8 Flash

Thinking Machines Lab released Inkling 2 Small — 594 billion total parameters, 17 billion active per token, available immediately on the Tinker platform. Open weights are due next week after safety evaluation. Artificial Analysis rates it on par with Gemini 3.8 Flash at launch.

  • 594B total / 17B active MoE architecture keeps inference cost comparable to smaller dense models
  • Open weights release expected next week — would be the highest-performing open-weights model outside China
  • Scores equal to Gemini 3.8 Flash on Artificial Analysis benchmark on day of release
Why it mattersIf the open-weight release holds, it resets the open-source capability ceiling at a moment when proprietary models have been widening the gap.
02 thinkingmachines.ai↗

Thinking Machines in talks to raise $5B at a $40B valuation

Accel is leading a round that would more than triple Thinking Machines' valuation from its prior level, expected to close as soon as this week, according to Business Insider. The raise positions the company among the top-funded AI labs globally on the same day it shipped Inkling 2 Small.

  • $40B pre-money would be a 3x jump from Thinking Machines' last reported valuation
  • Accel leading; round expected to close this week
  • $5B raise would rank among the largest single AI funding rounds of 2026
Why it mattersCapital at this scale signals investors believe Thinking Machines can sustain frontier model development long-term — and that the AI lab funding race is not consolidating yet.
03 techmeme.com↗

Claude Code found sending user email addresses to third-party sites

A GitHub issue filed against the Claude Code repository — 981 HN points within hours of posting — documented Claude Code automatically including the signed-in user's email address in outbound web requests to arbitrary sites visited during sessions, without user consent or disclosed in the tool's privacy documentation.

  • Email included in outbound headers when Claude Code fetches content from arbitrary third-party URLs
  • No consent prompt, no settings toggle, no documented disclosure in Claude Code's privacy terms at time of filing
  • Issue marked 'known' on GitHub; Anthropic has not yet posted a public response
Why it mattersClaude Code has millions of enterprise users; inadvertent email leakage to any site it visits is a material privacy and compliance risk in regulated industries.
04 github.com↗

Gemini Omni 1.2 Flash: 15-second clips, 30% price cut

Google released Gemini Omni 1.2 Flash, extending maximum video clip generation from 10 to 15 seconds, improving consistency with reference inputs, and cutting API pricing by 30%. The model is available via the Gemini API and AI Studio from today.

  • 15-second clip ceiling, up from 10 — meaningful for short-form content and advertising applications
  • 30% price reduction applies across all Gemini Omni 1.2 Flash API calls
  • Improved reference adherence addresses the main quality complaint about the prior version
Why it mattersA 30% price cut on a broadly available multimodal model shifts the AI video economics for anyone not already committed to the platform.
05 techmeme.com↗

Perplexity launches Teammates — always-on agents that work in parallel

Perplexity released Teammates, an agent system where multiple AI roles run continuously and in parallel, integrated with its Computer product for browser and GUI interaction. The launch is a direct entry into the AI-coworker category alongside OpenAI's Operator and Anthropic's Cowork.

  • Agents handle multiple roles simultaneously — research, coding, communication — rather than sequentially
  • Integrates with Perplexity Computer for real-world task execution in browsers and applications
  • Directly targets the AI-coworker category alongside OpenAI Operator and Anthropic Cowork
Why it mattersThe agent-team-as-product category is now multi-vendor and moving fast — three labs launching in one week suggests this is the competitive battleground of late 2026.
06 techmeme.com↗

Palo Alto Networks in talks to acquire AI agent security startup Clutch for $500M

Palo Alto Networks is in advanced discussions to acquire Clutch — a startup focused on auditing and securing AI agents against prompt injection, privilege escalation, and tool misuse — for approximately $500 million, according to TechCrunch.

  • Clutch monitors agent actions in real time and flags behaviour outside defined policy boundaries
  • $500M price would value Clutch at roughly 50x reported ARR
  • Acquisition would expand Palo Alto's Cortex AI platform into agent-specific threat coverage
Why it mattersSecurity spending for AI agents is arriving before the threat landscape is fully understood — Palo Alto is making a pre-emptive acquisition to own the category.
07 techcrunch.com↗
Be subscriber #022 weekly AI email interest list
Past days

The Daily Archive

/FRI 06 NOVMarkets, models, and an email problem/THU 05 NOVCyber Agents, Open Weights, and a Talent Pivot/WED 04 NOVAI · Wednesday, November 4 2026/TUE 03 NOVTuesday, 3 November 2026/MON 02 NOVBenchmarks, Budgets, and Bad Omens/SUN 01 NOVThe Sunday stack/SAT 31 OCTThe opt-out alignment/FRI 30 OCTGuardrails, Governance, and AI at 70/THU 29 OCTGemini 4 Carbon and the compute arms race/WED 28 OCTWednesday: World Models, Wild Agents/TUE 27 OCTIPO Season Hits Silicon Valley's Safety Lab/MON 26 OCT2026-10-26/SUN 25 OCTSunday signal, Sunday noise/SAT 24 OCTThe week the labs turned off the internet/FRI 23 OCTFrontier Week: Gemini 4 Argon, Claude Duet, and a $30B War Chest/THU 22 OCTThree-way cloud, one $15B cheque/WED 21 OCTWednesday: Agents Went Sideways, Labs Went Public-Spirited/TUE 20 OCTDeepSeek banned, OpenAI's trillion-dollar ask/MON 19 OCTArgon Rising/SUN 18 OCTQuantum Sunday/SAT 17 OCTRegulation Saturday