Accenture just “exited” more than 11,000 employees because they couldn't keep up with AI.

Meanwhile OpenAI is out here signing billion partnerships like they're collecting Pokemon cards.

Wild.

And that's not even the craziest part of the week. 

US President Trump posted then deleted an AI-generated Fox News clip promoting magical disease-curing medbeds. 

Just another wild week in AI, honestly.

Here's what matters, y’all. No fluff. 

What's happening with Accenture's workforce changes?

Reports show that Accenture reduced its workforce by nearly 12,000 people over the past three months as part of a nearly $1 billion AI-focused overhaul. The job cuts are targeting employees whose skills do not align with the company's AI-driven strategy.

The company is reinvesting those savings into AI training and plans to increase its overall headcount by 2026, focusing on AI and data specialists.

More layoffs are expected at Accenture as part of this ongoing restructuring. CEO Julie Sweet said the plan is moving on a compressed timeline and warned that further layoffs may follow if staff cannot adapt to an AI-first environment.

The company already spent $615 million on severance packages and restructuring with an additional $250 million expected this quarter. Accenture says retraining programs in cloud computing, data analytics, and AI are in place to redeploy talent.

Analysts note that this move aligns with a broader pattern across consulting and tech firms such as IBM, Microsoft, and Google as companies streamline teams and adopt AI-assisted workflows to boost productivity.

What it means: The big tech companies laid off tens of thousands first, then management consulting companies are following, and the rest of enterprise will likely follow suit.

If your company hasn't started AI upskilling programs, the window is closing.

Make sure your day-to-day business processes that rely on AI are easily transferable to new model methodologies if old models get phased out.

Wait, ChatGPT does what now?

OpenAI just introduced Pulse for pro subscribers on that $200 monthly plan. It works while you sleep to surface tailored daily updates based on your past chats and connected apps.

Sounds kinda creepy but also kinda useful.

The system uses information from your ChatGPT history, explicit feedback you share, and optional connected apps like calendars to curate suggestions. It generates card-based summaries each morning with actionable items like upcoming meetings, trip schedules, and research.

Updates disappear after a day unless you save them.

OpenAI emphasizes that Pulse is designed to reduce mindless scrolling rather than keep you glued to the app. Users can leave feedback to curate what Pulse recommends.

The company frames this as a step toward assistance that can research, plan, and take helpful actions on your behalf. Moving ChatGPT from reactive Q&A to proactive workflow tool.

Right now it's limited to paid pro subscribers, but OpenAI hinted this will roll out to all paid subscribers eventually. Maybe weeks, maybe months, maybe quarters.

Here's where it gets interesting though. The overall sentiment has been overwhelming. This is one of the most well-received ChatGPT features in recent memory.

What it means: 

ChatGPT just shifted from answering questions to proactively working on your behalf.

For the 99.9% of users who weren't already using hyper-personalized scheduled research, this will be genuinely useful.

The real play here is likely advertising, but for now it's helping people stay organized.

You're not gonna believe this one.

US President Donald Trump posted an AI-generated video styled as a Fox News broadcast claiming Americans could receive MedBed care, granting them access to hospitals with technology that restores every citizen to full health. (Seriously)

He posted it Saturday night on Truth Social.

Then deleted it Sunday.

Awkward.

Fox News confirmed the video was AI-generated and never actually aired on their platforms. The video claimed users would get a MedBed card with guaranteed access to top doctors and the most advanced technology, suggesting cures for any disease.

This is a hallmark of the MedBed conspiracy narrative circulating in fringe far right circles.

Look, regardless of your political leanings, we gotta talk about this. When the president of a leading country puts out clearly AI-generated content meant to deceive people, that's a problem.

People on Truth Social are legit selling MedBed cards following this video. Seniors and older Americans who see what appears to be a Fox News clip from the president might think it's real.

As AI video quality gets better, this type of disinformation will become commonplace.

What it means: 

High-quality AI video is now good enough to fool people when it comes from seemingly trusted sources.

Your company needs policies about verifying video content before acting on it, especially from social media.

Train your team to look for AI generation artifacts and always verify through official channels before believing viral clips.

NVIDIA and OpenAI casually dropped a $100 billion AI pact like it was a normal Thursday afternoon press release.

NVIDIA will invest $100 billion in a new OpenAI pact to build at least 10 gigawatts of AI data centers for the company behind ChatGPT, according to their joint announcement. Right now it's a letter of intent, not a finalized check.

But still.

The first one gigawatt of NVIDIA systems will be deployed in the second half of next year with full details to be finalized in coming weeks. This complements earlier commitments from Microsoft, Oracle, SoftBank, and Stargate partners who pledged at least $100 billion collectively in January.

One gigawatt is roughly the output of a large nuclear reactor, btw.

OpenAI CEO Sam Altman told CNBC the NVIDIA-built centers are in addition to previously announced projects. The infrastructure is essential to delivering services and improving models.

Altman's stated he wants to expand compute so much that within a year or two, OpenAI creates in one week what the entire industry currently produces in a year.

Wild ambitions require wild infrastructure investments, apparently.

What it means: OpenAI is going all-in on compute capacity to power next-generation models.

If your company is building on OpenAI's platform, this signals more reliability and less rate limiting ahead.

But AI compute costs will remain sky-high, so optimize your prompts and usage now.

Welp. More AI from Meta, we guess? Yikes.

Meta just announced Vibes, a new feed in the Meta AI app and on meta.ai for creating and sharing short-form AI-generated videos. Yup, they launched a full social media feed that's AI videos only.

Early reaction? People are calling it a trough instead of vibes.

Ouch.

The current feed isn't personalized yet and shows an eclectic mix of AI clips. Think an elephant driving a convertible, a Viking doing a selfie with Bon Jovi. Meta is seeding a notable portion of the content with company-created AI accounts.

This is part of their partnership with MidJourney and other AI media companies. Theoretically anyone can join Vibes and start creating videos.

Public reaction has been largely negative. Many users are calling the feed unnecessary and lacking purpose beyond showcasing AI's ability to generate random visuals. People are using the term "AI slop."

Here's the thing though. Six months ago, creating AI videos required actual human effort. Now anyone with one prompt can make an okay looking video.

What it means: 

AI video generation has become so easy that platforms can auto-fill entire feeds with it.

The concern is whether this devalues creativity and contributes to more doom scrolling.

Don't let your brand jump on every AI trend just because it's possible. Ask whether it actually adds value first.

How many billion dollar deals can one company announce in a week?

OpenAI announced plans to build five new US data centers that weren't previously announced under its Stargate initiative. This major scale-up will push total planned capacity to nearly seven gigawatts.

That's roughly the output of seven large nuclear reactors, y'all.

The expansion is being developed in partnership with Oracle and SoftBank. Construction is already underway at at least one of the new sites in Texas, with photographs from April showing active development.

Oracle's involvement points to deeper integration with its cloud services. SoftBank's participation suggests ongoing investments in AI infrastructure and ecosystems beyond chips.

Between this announcement and the NVIDIA pact, OpenAI had three different partnerships announced this week representing multiple billion dollar commitments with three different parties.

Absolutely bonkers week for OpenAI on the partnership front.

What it means: 

OpenAI is building infrastructure to support models that are orders of magnitude more powerful than what exists today.

If you're planning AI implementations for 2026 and beyond, factor in that model capabilities will likely explode.

Start thinking about use cases that seem impossible now but might be trivial in 18 months.

Yeah, we’ve heard this story like umpteen times.

Apple is reportedly testing an internal ChatGPT-style LLM called Veritas to prepare a major Siri overhaul, according to Bloomberg. The revamped assistant is currently slated for a March 2026 debut.

We've literally heard versions of this for three years straight though.

Veritas is used to trial a new system named Linwood at Apple, which powers the upgraded Siri. It combines Apple's own large language models with at least one third-party model.

Features under testing include AI search across personal data like your music library, emails, and messages. Plus in-app tools like AI-driven photo editing.

Apple is reportedly evaluating whether certain capabilities should live inside Siri or be a dedicated third-party chatbot. They're trying to figure out if Veritas should be a consumer product or just an internal tool.

Apple Intelligence launched last year using a hybrid approach of their own models and partner models, mainly ChatGPT. For anything complex, Apple Intelligence asks if you just wanna send it to ChatGPT.

Apple will also reportedly launch an AI-powered search engine called World Knowledge Answers integrated first in Siri then later in Safari in 2026. They've entered into an agreement with Google to use Gemini AI models as part of that system.

Similar announcements about major Siri improvements have been made repeatedly over the past few years.

What it means: 

Apple continues to announce Siri improvements while competitors move ahead with their AI assistants.

Don't hold your breath waiting for Apple to catch up. Plan your workflows around tools that work today.

If you're in the Apple ecosystem, accept you'll need to rely on third-party AI apps for serious work for at least another year.

Perplexity released an AI-powered email assistant but only for users on their $200 max plan. Meta rolled out an AI-powered dating assistant in what's turning into their week of wild swings.

Claude has landed in Microsoft 365 Copilot. Distyl AI raised $175 million at a $1.8 billion valuation. Suno launched v5 of their AI music maker and it sounds really good.

Meta is in talks with Google to use Gemini for Facebook ads instead of their own Llama products. OpenAI released GPT Bao, a new benchmark comparing how different models do for real economically valuable tasks.

Spotify cracked down and cut 75 million AI tracks from the platform. Google released Gemini Robotics 1.5. Moonshot released Okay Computer in agent mode for their popular Kimi model.

Microsoft unveiled a new marketplace with 3,000 AI apps and agents. OpenAI tapped CoreWeave for a $6.5 billion compute partnership. ChatGPT rolled out the ability to share projects for teams.

Perplexity released its search API to compete with Google. Google released an updated version of Gemini 2.5 Flash and Gemini 2.5 Flash-Lite. Extremely impressive updates, btw.

Microsoft is testing a marketplace paying publishers royalties for AI content use. ChatGPT confirmed they were routing some GPT-4o queries to a GPT-5 model tuned for safety. Those people who wanted GPT-4o forever weren't happy about that.

Notebook LM may start saving chat history without having to save them as notes. OpenAI rolled out parental controls for ChatGPT.

You catch all that? lolz.