OpenAI Is Bringing on Some Big Guns in the Lead-Up to Its IPO
Transformer co-inventor Noam Shazeer — the guy whose paper literally created the architecture underpinning modern AI — is leaving Google DeepMind for OpenAI, alongside former Trump AI policy official Dean Ball. Whether Shazeer is coming over to build the next thing or just to collect pre-IPO equity is a question worth asking, but either way this is a significant get. OpenAI clearly wants to walk into its IPO roadshow with the most impressive possible roster, and “we hired one of the people who invented transformers” is a pretty good slide.
Barret Zoph Is Out at OpenAI Again After Just Five Months
Five months. The man left OpenAI to co-found a competing company with Mira Murati, returned to OpenAI in January to run enterprise AI sales, and has now departed again — which means he’s done the full exit-return-exit cycle in roughly the time it takes to vest a single stock tranche. At this point Barret Zoph’s LinkedIn must read like a revolving door instruction manual. Whatever’s happening inside OpenAI’s enterprise division, the churn at the top is not inspiring confidence.
The White House Is Making Up Its Rules for AI in Real Time
Wired’s deep dive confirms what’s been maddening about the Anthropic/Fable 5 situation: nobody — not Anthropic, not lawyers, not apparently anyone in the administration — can articulate exactly what rule was broken. The Trump White House is effectively improvising AI governance on the fly, issuing blocking orders without clear statutory grounding. This is less “regulation” and more “vibes-based authority,” which is a genuinely alarming way to run policy for one of the most consequential technologies in human history.
Amazon Workers Say They’re Facing Termination for Backing Data Center Limits
Three Amazon software engineers testified at Seattle City Council hearings about data center expansion, citing a city ordinance specifically protecting political speech from employer retaliation — and then, one week later, found themselves under internal investigation. Amazon’s legal exposure here seems obvious, which makes the alleged retaliation either arrogant, stupid, or both. The broader subtext: as AI infrastructure buildout accelerates, the people actually building it are starting to push back, and their employers are not handling it gracefully.
Photoshop and Premiere Now Have AI Assistants
Adobe is rolling out bespoke AI assistants across Photoshop, Premiere, Illustrator, InDesign, and Frame.io simultaneously, which is either a sign of genuine product maturity or a sign that Adobe’s competitive anxiety about Figma and every AI-native creative tool has reached critical mass. The Firefly studio update — which now remembers your previous work and maintains context across projects — is the more interesting piece, since persistent memory is where these tools go from “neat demo” to “actually changes how I work.”
OpenAI Is Improving Health Intelligence in ChatGPT
GPT-5.5 Instant is rolling out with specifically tuned health and wellness responses, developed with physician input and evaluated against clinical communication standards. Combined with the separate announcement that an OpenAI reasoning model helped physicians identify 18 new rare disease diagnoses in previously unsolved cases, there’s a clear pattern here: OpenAI is making a serious, sustained push into healthcare. The question isn’t whether AI can be useful in medicine — it clearly can — the question is what the liability and accountability framework looks like when it gets something catastrophically wrong.
If AI Is Sentient Then So Is ‘Age of Empires II’
A new paper makes the formal argument that the criteria people use to claim AI sentience would, applied consistently, also confer sentience on a 1999 real-time strategy game. “The point of the paper is to formally show that we anthropomorphise too readily,” the researchers note — which is doing a lot of heavy lifting as a sentence. This is genuinely useful philosophy dressed up in a funny headline, and it’s a needed corrective to the cottage industry of AI consciousness discourse that has bloomed around every chatbot that says something unexpectedly poetic.
AI Inference Startup Baseten Reportedly Raising $1.5B Months After Its Last Mega-Round
Baseten is reportedly closing a $1.5 billion round at a $13 billion valuation — and the notable detail is that this comes just months after their previous massive fundraise. The inference layer of the AI stack is where the real infrastructure battle is playing out right now, and investors are clearly not waiting for consolidation before writing enormous checks. Whether this level of capital formation into inference infrastructure is justified by actual demand or is the frothiest part of the current bubble is, charitably, an open question.
Bottom Line
The AI industry this week is a perfect microcosm: brilliant hires and record fundraises on one side, governance chaos and employee retaliation on the other — the technology is moving fast, and literally everything around it is still making it up.