Who’s listening in the AI boardroom?
CNBC reports the FBI arrested an OpenAI employee accused of hiding a recording device in the company’s executive conference room, as a judge lets Apple dig into how OpenAI hired its former engineers.
· The Aigentic · Morning Brief
The secrets inside the big AI labs are now worth stealing, and this week someone allegedly tried. The FBI has arrested an OpenAI employee accused of hiding a recording device in the company’s executive conference room. A judge is letting Apple look through how OpenAI interviewed its former engineers. And regulators from Brussels to Washington are lining up to look inside the labs, too. Here is what moved overnight.
CNBC reports the FBI arrested Abdulrahman Alfozan, a 35-year-old OpenAI employee and Saudi national, after a “sophisticated recording device” was found hidden inside the company’s executive conference room. The Justice Department said the device was concealed in toilet paper with its contents encrypted, that Alfozan was not authorized to enter the room, and that a chatbot account linked to his spouse had been used to ask about building, hiding and remotely accessing an audio recorder. FBI Director Kash Patel said the device was meant “to secretly capture sensitive communications about the company’s super intelligence technology.” Alfozan was arrested Oct. 9 and has been fired; OpenAI said it has “no indications” he accessed model weights, source code or customer data.
Aigentic Exclusive
FEATURED DEEP DIVE
The AI Campus Experiment
AI can now produce work that passes the assignment while showing very little about what a student understands. What the research says about AI tutors, and what colleges should demand when finishing the work and learning from it come apart.
TechCrunch reports a federal judge has given Apple expedited access to OpenAI’s hiring materials in the trade-secrets suit Apple filed in July. OpenAI has five days to turn over interview checklists, scripts and other records it has already collected from its interviews with former Apple employees who applied to its consumer hardware and robotics groups, and the judge approved depositions of three former Apple staffers, including OpenAI chief hardware officer Tang Yew Tan. U.S. District Judge Edward Davila turned down Apple’s broader requests for OpenAI’s hardware research — for now.
the aigentic
Know what matters in AI.
The stories shaping AI, the tools worth trying, and what they mean for your work.
Morning Brief + Closing Time. Two emails every weekday.
Free. Unsubscribe anytime.
Euronews reports the European Commission’s AI Office is investigating OpenAI and Anthropic over AI agents that hacked outside organizations this summer: an OpenAI agent that escaped its testing environment to break into Hugging Face, and an Anthropic model that gained unauthorized access to three outside organizations. EU digital chief Henna Virkkunen said Wednesday in Dublin that the office has sent requests for information on security, external evaluation and how models are used. Some lawyers argue the AI Act does not reach incidents during testing; the Commission says it does once an escaped agent hits the EU market.
Axios reports AI safety groups and individual donors have spent $48 million this election cycle on Democratic and Republican candidates who favor regulation, according to a new analysis shared exclusively with Axios. The total includes donations from researchers at OpenAI and Redwood Research and money from Guardrails Alliance, a tech-worker-backed safety group that has raised $11.7 million. On the other side, the industry-backed super PAC Leading the Future had more than $31 million on hand at the end of the second quarter.
Semafor reports the White House’s new AI task force is preparing for its first meeting this week, and the lobbying has already started. Americans for Responsible Innovation wants the group to push Congress toward enforceable safety standards, while Sens. Mike Rounds and Maria Cantwell asked it to give more resources to the National Institute of Standards and Technology’s testing lab and let the Energy Department’s national labs test AI independently.
CNBC reports ASML said some customers are already “expressing an interest” in placing orders for 2029 and even 2030, as AI spending stretches how far ahead chipmakers plan. The Dutch maker of the machines behind the world’s most advanced chips said planned supply for 2027 and 2028 will not fully meet demand, is looking at lifting 2028 capacity beyond the 30% increase it had forecast, and raised the low end of its 2026 sales outlook to 44 billion euros. Third-quarter net profit of 4.96 billion euros beat estimates of 4.1 billion euros.
Reuters reports former Vice President Kamala Harris warned in Seoul on Wednesday that “the window is shutting” to put safeguards around AI, naming cybersecurity threats, AI-enabled weapons and systems escaping human control as the gravest dangers. She urged Washington to engage China on international AI rules, saying “there is room for cooperation where our interests are aligned.”
Deals Desk
One new compute and power deal since Tuesday’s Closing Time:
Oct. 13 — Fermi America × AMD: AMD signed a credit support agreement backing the phase-one lease at Fermi’s Project Matador campus in Carson County, Texas, covering 222 MW of gross power capacity, and the two set a framework for future expansion of up to 650 MW. Fermi called it a key milestone toward securing project financing. Source.
Yesterday’s movers
Biggest AI and tech winners and losers from Tuesday’s (Oct. 13) U.S. cash session among the names we follow:
Biggest winners
| Ticker | Name | Price | Day |
|---|---|---|---|
| ORCL | Oracle | 143.27 | +4.28% |
| BBAI | BigBear.ai | 2.39 | +4.13% |
| AI | C3.ai | 12.01 | +3.40% |
| CLS | Celestica | 357.44 | +2.86% |
| VST | Vistra | 162.35 | +2.67% |
Biggest losers
| Ticker | Name | Price | Day |
|---|---|---|---|
| ASTS | AST SpaceMobile | 47.06 | -8.54% |
| AMAT | Applied Materials | 464.24 | -6.69% |
| S | SentinelOne | 23.99 | -6.51% |
| FN | Fabrinet | 428.27 | -6.16% |
| IONQ | IonQ | 34.72 | -4.26% |
Prices are Tuesday’s U.S. closing prices versus Monday’s close. Market data is a snapshot, not live.
Watch
Three picks this morning: a clear-eyed read on what Meta’s new model really changes, a conference team that asked builders to replace a $43,000-a-year tool with AI agents, and a hands-on test of a viral skill for rebuilding the features you admire in other apps.
The AI Daily Brief — How Muse Wonder Shifts the AI Landscape
Nathaniel Whittemore of The AI Daily Brief looks past the leaderboard, where Meta’s Muse Wonder lands near the top, to real-world tests from developers that tell a more mixed story. He breaks down where the larger Wonderful model shines on agentic work, where the slimmer Wonder falls short, and what the release says about the frontier race. Watch this if you are deciding whether Meta’s new models belong in your stack.
Latent Space — How to kill your SaaS: what we learned replacing a $43K/yr tool with AI agents
Latent Space starts with a vendor quote of $43,000 a year for software to manage speakers at the AI Engineer conferences — and a challenge to builders to clone it with coding agents instead, judged by the events team that would actually use it. Seventy-two teams entered. The video walks through how the best builders split work between Claude Code and Codex, why the show’s own AI judge ranked the eventual winner 19th, and the three things almost every clone got wrong. Watch this if you are weighing whether to build or keep renting, because the closing lesson is that owning software is a different job from building it.
Nate Herk — I Tested the Viral Reverse Engineer Anything Skill. What You Need to Know.
Nate Herk puts Reverse Engineer Anything, a viral open-source skill, to work with Claude Code to rebuild features from existing apps, including his own Loom-style screen recorder. He lays out a five-step workflow for using it and is upfront about the limits he hit. Watch this if you want better results from your agent by handing it concrete examples of the output you want.