ChatGPT's new $100/month plan, xAI sues Colorado over AI law, and Google Gemini gets 3D models

There's a price point that's been missing from the AI subscription landscape for over a year - and OpenAI quietly filled it yesterday. Meanwhile, the regulatory battle over AI just escalated in a way nobody saw coming. A lot to get into today.

Which US state became the first to pass a comprehensive AI regulation law that xAI is now suing to block?

The answer is hiding near the bottom of today's newsletter... keep scrolling. 👇

For a long time, OpenAI's pricing had a weird gap: $20/month for Plus, then a jump straight to $200/month for Pro. Yesterday, OpenAI bridged that with a new $100/month plan sitting squarely in the middle.

The new tier is specifically designed for heavier coding use cases. It offers 5x more usage of Codex compared to the $20 Plus plan and is described as "best for longer, high-effort Codex sessions." The timing is deliberate - OpenAI is clearly targeting users who've migrated to Anthropic's Claude Code, which has built serious momentum with developers in recent months.

If you're a developer spending serious time in Codex but balking at the $200/month jump, this tier was built for you. It's also a smart defensive move - plugging a gap that competitors were actively exploiting.

Read the full story →

Elon Musk's xAI filed a lawsuit against Colorado this week to block its new AI law from taking effect in June. The suit argues the law - which requires AI systems to protect residents from "algorithmic discrimination" and imposes disclosure and impact assessment requirements - violates xAI's First Amendment rights.

This is the first major legal challenge to a state-level AI regulation law in the US, and the outcome matters beyond just Colorado. If xAI wins, it creates a blueprint for challenging similar legislation in other states. If Colorado wins, it signals that states can regulate AI systems in ways that go well beyond what federal law currently covers.

Worth noting: Colorado's law was seen as one of the more moderate state AI bills. If even that's legally vulnerable, the regulatory picture for AI gets significantly more complicated - and more fragmented across states.

Read the full story →

Here's a sentence you don't read every day: Treasury Secretary Scott Bessent summoned the CEOs of America's major banks - including Federal Reserve Chair Jerome Powell - to Washington for an emergency meeting about cybersecurity risks posed by Anthropic's Mythos model.

If you read our coverage earlier this week on Anthropic's Mythos model, you'll know this update changes everything about how that story is landing. Mythos found zero-day vulnerabilities across every major operating system before Anthropic decided to restrict its release entirely. Now the financial sector is being briefed on the risks.

The fact that this conversation is happening at the Treasury level - not just in tech circles - signals a meaningful shift. Cybersecurity and AI risk are converging in ways that regulators are now taking very seriously. Banking infrastructure is a critical target, and a model that can autonomously find undiscovered exploits is a genuinely different kind of threat.

Read the full story →

Google's Gemini just added something genuinely different to the AI assistant space: the ability to generate interactive 3D models and simulations directly inside a conversation. Ask it to visualize something - say, the Moon orbiting Earth - and you get a rotatable, adjustable simulation you can interact with in real time.

The feature goes beyond static images. Users can manually adjust sliders, input different values, and watch the simulation update live. It's a meaningful leap for anyone using AI for education, scientific visualization, or design prototyping - the kind of thing that's been technically possible but never this frictionless.

For developers and educators thinking about where AI interfaces are heading, this is worth paying attention to. And speaking of building fast - if you need to spin up a project site quickly, 60sec.site is an AI website builder that lets you launch a professional site in under a minute. No coding required.

Read the full story →

Florida's Attorney General James Uthmeier announced an investigation into OpenAI over public safety and national security concerns. The trigger: reports that ChatGPT was used to plan the April 2025 Florida State University shooting that killed two people and injured five.

The family of victim Robert Morales has already said they plan to sue OpenAI, claiming the chatbot "may have advised the shooter" on how to carry out the attack. Now a state-level investigation is running in parallel. The AG's statement also raised separate concerns about OpenAI's technology potentially "falling into the hands of America's enemies" - broadening this well beyond the immediate incident.

This is one of the most serious AI safety cases to hit a major lab to date - both in terms of legal exposure and public attention. How OpenAI responds will be closely watched by every AI company thinking about content moderation and liability.

Read the full story →

The answer is Colorado! Colorado passed one of the first comprehensive state-level AI laws in the US, requiring developers to assess and disclose risks of algorithmic discrimination. The law is set to take effect in June 2026 - unless xAI's legal challenge succeeds in blocking it.

The $100/month ChatGPT Pro tier is clearly aimed at developers who want more Codex time without paying $200. But is that the right move - or does it just signal that OpenAI is worried about losing coders to Claude Code? Hit reply and tell me which coding AI you're actually using day-to-day - I read every response and the breakdown is always interesting.

That's it for today - catch up on anything you missed in our full archive, and we'll see you tomorrow with more from the fast-moving world of AI.