Perplexity pivoted hard and is going after Siri. 

The U.S. government pivoted hard and is going after AI literacy in schools. 

And OpenAI is also pivoting hard to reportedly release an ‘open’ model by this summer. 

So many pivots, so little time. 

Don’t waste hours a week trying to keep up with what AI news means for you or your company. We do that for you on Mondays with our AI news that Matters segment. 

So…. let’s cut the B.S. and marketing fluff and see what’s new and why it matters. 👇

OpenAI released a lightweight version of its deep research tool for all users, including free accounts. Free users get 5 monthly queries, paid users get 25, and Pro subscribers ($200/month) enjoy 250 premium searches before downgrading.

The lightweight version surprisingly runs on o4 mini instead of o3. Yes, the "lighter" version uses the newer model architecture – AI logic at its finest.

OpenAI and Google remain the undisputed research kings, with Perplexity fading in their rearview mirror.

What it means: 

Information gatekeeping just died a spectacular death. Pew pew. 

The research playing field is flattening faster than anyone predicted.

Today's free tools demolish what premium subscribers paid hundreds for last year. By 2026, middle schoolers will conduct PhD-level research between TikTok sessions. The knowledge gap? Dissolving faster than your New Year’s Resolutions by Jan 7. 

A federal judge declared Google a search monopoly, and now the DOJ wants Chrome sold off. The browser commands 66% global market share, and – plot twist – OpenAI already wants to buy it.

The DOJ also wants Google sharing user data with competitors, potentially killing their $20B Apple deal, and stripping Google apps from Android. Google is fighting back hard.

Court docs revealed Gemini has 350M monthly users versus ChatGPT's 600M, with Altman hinting weekly users might actually hit 1B.  

What it means: 

This DOJ plan creates a different monopoly wearing nicer clothes. Microsoft already built Edge on Chromium – that's literally the point of open source. 

American regulators are playing checkers while Chinese tech plays quantum chess. Also, if OpenAI were to somehow acquire the Chrome Browser, wouldn't that just give them a Monopoly? 

Make it make sense. 

Microsoft launched "wave two" Copilot updates with an entire agent store for research and analysis. 

Their killer feature in an age of AI everywhere and humans nowhere? A skill discovery agent that helps humans find other humans with specific expertise – AI promoting actual human connection! Score one point for humans. 

Other goodies include brand-compliant image generation, Notebooks combining all work formats into one searchable space, and natural language search across Microsoft apps and third-party tools.

What it means: Microsoft keeps packing in the AI features like trying to eat more food without changing the tight pants. 

Organizations using Microsoft's ecosystem have the continued chance to be ultra efficient by Q4. That is… if you and your IT team know where all the Copilot features are and how to enable them. 

Then train your team on them. 

…..

And keep up with all these updates. Lolz. 

Adobe shocked everyone by inviting OpenAI and Google models into its Firefly AI platform. Users will soon be able to generate with OpenAI's GPT Image, Google's Imagen 3, and various video models, with more integrations coming soon.

Create with any model, edit in Photoshop. Seamless. Adobe stock jumped 2% immediately – smart money recognizes a survival play could actually be a power growth move. 

What it means: 

Adobe finally faced reality. Google AI Studio is basically free Photoshop now, and OpenAI's images are churning out Adobe-worthy visuals with nothing more than a text prompt. 

Adob’s play? Become the Switzerland of AI imagery. If executed well, Adobe transforms from product company to essential platform. Smart creators will use Adobe as home base while cherry-picking the perfect model for each specific task.

Mike Lindell's lawyer submitted 30 FAKE legal citations to a federal judge – all AI-generated fiction. The attorney admitted using AI to "finalize" without verification, and Judge Nina Wang gave them until May 5 to explain why they should keep their licenses.

Rut roh. 

MyPillow's already reportedly down $100M in revenue after retailers fled post-election conspiracy theories. This AI-generated legal citations fiasco only compounds their troubles.

What it means: 

This case will become the legal AI horror story professors tell first-years for the next decade. That 2023 horror story of the New York lawyer using ChatGPT to create false citations will now just be an afterthought for this My Pillow Snafu. 

The problem wasn't the AI – it was lazy humans skipping verification. And not knowing how AI works. 

Large language models rarely hallucinate with proper human expertise and if you’re using the right tool for the right job.

 Bar associations should probably mandate AI certification by 2026, and malpractice insurance rates for lawyers will skyrocket faster than crypto in a bull market.

And just like that…. The creative industry is completely changing. 

OpenAI released their image generation API after 30M users created 700M images in one week. Developers can generate multiple images simultaneously with adjustable quality settings and safety measures.

All images include C2PA watermarks identifying them as AI-generated – from consumer toy to developer ecosystem overnight.

What it means: 

Creative app explosion incoming.

 The real innovation won't be in general image creation but in solving specific visual problems for industries.

Visual content production costs will crater as AI-powered tools make in minutes what previously took weeks. 

The only thing holding your creative ambitions back now is reps. 

U.S. President Trump signed an executive order boosting AI education nationwide with a White House Task Force. The initiative prioritizes early AI training with a Presidential Challenge showcasing student achievements.

One hiccup: The Department of Education budget was slashed earlier this year. Implementation details remain conspicuously absent.

What it means: 

America finally woke up to the AI education emergency – our schools are woefully unprepared. Many still BANNING the technology students need to master!

The tricky part? 

This will require the now-depleted U.S. Department of Education to pull this off. You know…. The one whose head kept calling Artificial Intelligence ‘A1” like the steak sauce. 

Yikes.

Microsoft is finally/really/actually rolling out its controversial AI Recall feature after 10 months of security overhauls. The feature screenshots everything you do and makes it searchable – now with encrypted storage and sensitive info filtering after security experts raised alarms.

Windows search now supports natural language queries, and a new "Click to Do" feature lets you summarize text or edit images with one keystroke + click.

What it means: 

Apple's search system just became a fossil. Finding files on a Mac without exact names? Pure torture.

 Knowledge workers will save hundreds of hours yearly with "show me that document with the red chart from last week" actually working.

The productivity gap between Windows and Mac just became significant enough to influence purchasing decisions for entire organizations.

Perplexity launched a voice assistant on iOS that makes Siri look prehistoric. It creates calendar events, sets reminders, plays music, and holds actual conversations with context memory. Well…. Sometimes. 

The only problem? Consistency. Works brilliantly once, fails the next three tries. Fix that, and they've got something huge.

What it means: 

Big pivot for Perplexity. 

We called in January that Perplexity would have to pivot or get squared. 

Perplexity just made their only viable play after OpenAI and Google crushed them at their own research game. Apple's neglect of Siri (delayed upgrades until 2027!) created this opening.

If they fix reliability issues fast, they could capture massive market share before Apple wakes up. Voice remains the most natural human-computer interface – whoever solves consistency wins big.

OpenAI will reportedly release their first open model since 2019 this summer – a text-only reasoning model for high-end consumer hardware with toggleable reasoning capability.

Sam Altman acknowledged they need an open source strategy after letting Meta and others dominate the space. Thorough safety testing and detailed model cards promised before release.

What it means: 

The open model landscape is about to get nuked from orbit if OpenAI actually does release an open source reasoning model. 

When the company with "Open" in its name finally delivers open code? 

Revolution. 

Techniques will filter into every AI system, accelerating development beyond what anyone thinks possible.