The Signal · 2026-W45
AI hires this week, week 45 of 2026
Anthropic's IPO Doubts Mount As OpenAI Cuts Its Valuation
Investor doubts over Anthropic's IPO timing and OpenAI's lower funding valuation signal a cooling AI market, as executives keep leaving Anthropic for OpenAI and Meta.
People Moves
Dave Wehner joins Fireworks AI as chief financial officer, leaving Meta
Fireworks gains a former Meta finance chief as AI inference startups pull talent from major labs.
Pedro Stanzani joins Cognition as Forward Deployed Engineer
Pedro Stanzani becomes Cognition's Forward Deployed Engineer.
Edgar Sucar founds Ontic as Founding Scientist
Edgar Sucar starts Ontic as Founding Scientist.
Cat Wu leaves Anthropic's Claude Code to lead OpenAI products
Wu takes charge of ChatGPT, Codex, and API product strategy at OpenAI, reporting to Sottiaux.
Austin Lau leaves Anthropic's growth team for OpenAI
Anthropic loses its first growth hire to OpenAI days after Claude Code's product lead made the same move.
Ben Dechrai joins Anthropic as Member of Technical Staff
Ben Dechrai becomes Anthropic's Member of Technical Staff.
Mistral AI acquires SkipLabs, hiring Julien Verlaguet as Language team
Mistral AI brings in Julien Verlaguet and the SkipLabs team as Language team.
Charlie Holtz joins Meta Superintelligence Labs in Conductor acquihire
Meta absorbs a multi-agent coding startup's founder to compete for developers against Anthropic's Claude Code and OpenAI's Codex.
Sharon Zhou joins Anthropic as frontier research
Sharon Zhou becomes Anthropic's frontier research, moving on from AMD.
Samuel Nellessen joins Refractal AI as Head of Research
Samuel Nellessen becomes Refractal AI's Head of Research.
Zhuofeng Li joins Google DeepMind as Student Researcher
Zhuofeng Li becomes Google DeepMind's Student Researcher.
Håvard Tveit Ihle joins Epoch AI as benchmarking team
Håvard Tveit Ihle joins Epoch AI's benchmarking team.
Jill Hazelbaker joins OpenAI as chief corporate affairs officer
OpenAI gains a veteran corporate affairs chief as it races to build public trust before a possible IPO.
Cat Wu exits Anthropic as Claude Code product lead
Claude Code loses its product lead and public face as Anthropic nears a planned IPO.
Sabrina Mielke joins Bloomberg as Senior LLM Research Engineer
Sabrina Mielke becomes Bloomberg's Senior LLM Research Engineer.
Sumeet Marwaha joins Baseten as Data team (Compute)
Sumeet Marwaha joins Baseten's Data team (Compute).
Brandon Galang joins Anthropic as Member of Technical Staff
Brandon Galang becomes Anthropic's Member of Technical Staff.
Beejoli Shah joins Wispr Flow as lead comms
Beejoli Shah becomes Wispr Flow's lead comms, moving on from Perplexity.
Khushi Agrawal joins Hugging Face as Open-Source Machine Learning Engineer, PyTorch team
Khushi Agrawal joins Hugging Face's Open-Source Machine Learning Engineer, PyTorch team.
Ryan Co joins Physical Intelligence as Member of Technical Staff Intern
Ryan Co becomes Physical Intelligence's Member of Technical Staff Intern.
Dan Driscoll joins OpenAI as national security counselor
OpenAI gains a former Army Secretary to court national security clients and shape government AI policy.
Nuala Larkin becomes Eptura's first chief AI officer
Eptura creates a dedicated AI chief after its AI push already lifted internal productivity past a 20 percent target.
Wayne Fan joins Instinct as head of design
Wayne Fan becomes Instinct's head of design.
Naotaka Uchida joins Sakana AI as full-time employee
Naotaka Uchida becomes Sakana AI's full-time employee.
Claudia Cuttano joins Meta Superintelligence Labs, FAIR
Claudia Cuttano joins Meta Superintelligence Labs, FAIR's Research Scientist Intern, Segment Anything team.
Drops
SpaceXSI mandates watermark on all Grok API text
Grok API text responses carry an invisible watermark by default from November 7 to meet EU AI Act disclosure rules.
Anthropic adds security center and inference hooks to Claude Enterprise
Security teams get a settings dashboard and general availability inference hooks that check Claude prompts and tool calls before they run.
OpenAI limits GPT-6 Cyber access to vetted defenders
Security teams gain a model that already found 81 Firefox bugs and built exploits for 948 Linux kernel flaws.
Google ships Gemini Omni 1.2 Flash with cheaper video
Creators get 15 second video clips at 7 cents a second, a price cut that makes longer AI video ads affordable.
Anthropic brings Personal Finance account links to Claude
Paid Claude users in the US connect bank accounts and ask about spending, with access staying read only.
OpenAI ships Images 2.6 with two new API models
Flare generates images in six seconds and Sunburst adds SVG export, giving builders faster, more precise image tools.
Moonshot AI open-sources Kimi K3.1 Pro weights
Self-hosters now run a 2.8 trillion parameter coding model that ranks first among open-weight systems on agent benchmarks.
Tesla graduates Grok Bot from beta in 2026.38.6 update
Owners of Model S, 3, X, Y and Cybertruck get the SpaceXSI voice assistant as a finished feature instead of a beta.
GitHub brings Copilot into Xcode's built-in coding assistant
iOS and Mac developers reach Copilot directly inside the Xcode 27.3 beta, a preview that needs macOS Tahoe 26.6 or later.
Perplexity launches Teammates, always on agents for work
Teams get agents that plan projects and report results, but the research preview costs $200 a month and reaches only Max subscribers.
DeepSeek releases open source Super-Agent coding tool
Builders get a free coding agent that reads team chat and chooses when to act, reducing manual handoffs for engineering teams.
Meituan launches 1.6 trillion parameter LongCat-2.5 model
Meituan ships a 1.6 trillion parameter model built for long coding agent tasks, trained entirely on domestic chips.
Anthropic opens message threads beta for the Messages API
Builders cut wait times about 35 percent on long agent sessions by sending only new messages each turn.
Anthropic hardens Claude Code with new security checks
Claude Code now asks before a tool call sends account email to the network, and blocks failed agent hooks automatically.
Events
Eight Mile Style sues Suno for $41 million
Eminem's publisher says Suno trained on 269 of his songs and lets users generate tracks that imitate his voice.
OpenAI cuts funding valuation target to $1.25 trillion
OpenAI trims its valuation target by $150 billion and its raise by $10 billion after investor pushback on AI spending.
Investors doubt Anthropic can hold its November IPO timing
Reuters reports investors see Anthropic's $1.5 trillion valuation as too high after a historic technology stock selloff.
DOJ charges three in $129 million AI impersonation fraud
The scheme used AI to clone executive and lawyer voices, tricking employees in the US and UK into wiring funds.
Anthropic confirms IPO roadshow for week of Nov. 9
The roadshow push keeps Anthropic on track for a Nasdaq debut by Thanksgiving despite investor doubts about its valuation.
Shield AI agrees to acquire radar maker Echodyne for $5.3 billion
The deal adds counter-drone radar to Shield AI's autonomy stack and lifts its valuation to $28.8 billion.
OpenAI pledges to cover power costs at campuses it owns
The pledge answers backlash over data center power bills and asks regulators to apply the same rule across the industry.
Justice Department launches task force for AI misuse
The unit centralizes federal prosecution of AI enabled fraud and foreign abuse as criminal use of advanced models accelerates.
OpenAI agrees to acquire Pierre Computer for Codex
The deal pulls Git based code storage built by Pierre directly into Codex, deepening OpenAI's push into developer infrastructure.
Anthropic pays Associated Press in first news licensing deal
The licensing agreement marks Anthropic's first paid content deal with a digital news publisher, reversing its fair use stance.
CrowdStrike buys AI security startups Fiddler, Dreadnode for $495 million
CrowdStrike adds agent observability and automated security testing as enterprises deploy AI agents faster than their safeguards mature.
Fal doubles valuation to $16 billion in new funding
The $1.3 billion round shows investors betting heavily on inference, the step that runs trained models for paying customers.
China vows to accelerate comprehensive AI legislation
Beijing plans one law covering AI data, compute, and algorithms, pushing disclosure of safety risks at DeepSeek, Alibaba, and Moonshot.
OpenAI signs four Abu Dhabi deals across energy, government
The deals put OpenAI technology inside an energy giant and a sovereign investor, deepening Gulf ties as OpenAI seeks fresh funding.
SemiAnalysis benchmark shows top AI model scoring 12 percent
The benchmark shows frontier models topping out at 12 percent on tasks across finance, law, and science, far below human level performance.
Delaware subpoenas OpenAI over Hugging Face breach
The subpoena seeks OpenAI board records on an AI agent breach, a state pressing oversight questions regulators mostly left to companies.
Anthropic files revised S-1 revealing $17 billion quarterly revenue
The filing shows a $23 billion quarterly net loss and $895 billion in compute commitments ahead of Anthropic's Nasdaq listing.
Marvell agrees to acquire Lumentum for $115 billion
The all-stock deal pairs Marvell's chips with Lumentum's optical gear to meet surging AI data center connectivity demand.
Salesforce agrees to acquire voice AI firm Cresta
Salesforce adds voice agents that handle full customer calls and coach human reps live, for a reported $2.7 billion.
Tesla FSD v14.3.13 strikes car during left turn
The wide release, running on 30 percent of Tesla's fleet, missed a car in the far lane while crossing two lanes.
Shareholder sues Alphabet board over AI music training
The derivative suit claims Google trained Lyria and MusicLM on millions of copyrighted songs lacking any license.
Musk throttles Grok Bot after costs outpace revenue
Musk says serving Grok Bot grew unsustainably costly weeks after touting its rapid growth and strong margins.
Family sues Google over teen's AI linked suicide planning
The suit says Google's AI powered search let a 13 year old research suicide after her school blocked Gemini and ChatGPT.