AI News Today – September 2

AI is getting more capable, but today’s biggest launch shows that even the companies building it are not comfortable giving everyone the same level of access.

Anthropic has released two new Claude models.

One is available to everyone.

The other is being kept behind a gate because its abilities in cybersecurity and biology could be misused.

There is plenty more happening across AI today, but we have picked out the biggest stories worth your attention.

Here is the AI news you need to know today.

Anthropic Claude Fable 5.1: Why Is Mythos 5.1 Restricted?

Anthropic’s newest Claude release comes with an unusual split.

Claude Fable 5.1 is now widely available, while Claude Mythos 5.1 is limited to vetted cybersecurity and life-sciences organizations.

They are essentially the same underlying model, but approved experts using Mythos get fewer restrictions.

Anthropic says Mythos 5.1 has the strongest cybersecurity abilities of any model it has released.

Its biology skills are also advanced enough that Anthropic specifically tested whether the model could help with dangerous chemical or biological work.

For everyone else, Fable 5.1 brings stronger research, coding and long-running work, while Anthropic says typical usage should cost about 25% less than Fable 5.

That is what makes this launch important.

AI is becoming useful enough for serious scientific and security work, but capable enough that its creators are starting to decide who gets access to its full abilities.

NYC AI Ban: Why Are 600,000 Students Losing Access?

Schools have spent the past few years trying to work out whether AI should be embraced or restricted.

New York City has just drawn one of the clearest lines yet.

The city is imposing a one-year ban on student-facing generative AI from preschool through eighth grade, affecting nearly 600,000 children in America’s largest public-school system.

Companion chatbots are also banned across every grade, teachers cannot use AI to grade work, and younger students face tighter screen-time limits.

High school students will still get limited supervised access and lessons explaining how AI works and where it can go wrong.

New York calls it the broadest student-facing AI moratorium in the country.

If other school systems follow, the debate may shift from how quickly children should learn AI to a much harder question: how young is too young?

Nvidia AI Regulation: Why Is Jensen Huang Telling the G20 to Back Off?

The AI race is becoming a battle over rules as much as technology.

At a G20 meeting today, Nvidia CEO Jensen Huang urged governments not to regulate AI around “theoretical harms” and instead focus on problems already happening in the real world.

The G20 brings together many of the world’s largest economies, giving these discussions global weight.

The United States is also pushing for clearer rules allowing AI companies to train systems on copyrighted books, articles and other creative work.

That question has already triggered lawsuits involving OpenAI, Anthropic, Google and Meta.

The eventual rules could decide what AI companies are allowed to learn from, how expensive new models become and whether creators must be paid or asked for permission.

The Rules Around AI Are Starting to Matter as Much as the Technology

Today’s G20 debate shows how quickly AI is moving beyond the companies building it.

Schools, governments and entire industries are now having to make decisions around technologies that barely existed a few years ago.

And AI is only one part of a much bigger shift.

The $5 Trillion Signal is our free, beginner-friendly briefing on five technologies we believe could reshape the global economy: AI, robotics, energy, biotech and blockchain.

It takes just a few minutes to read.

Enter your email below and we will send it straight to your inbox.