Google announced on 19 August 2026 that eligible US college students can claim a full year of Google AI Pro free of charge, normally US$19.99 a month, alongside a new student hub with study notebooks, 3D lesson visualisations, deep-research tools, and Gemini Live for talking through coursework.
It's the latest move in a pattern Google has run repeatedly this year: subsidise AI access hard at the exact moment habits are forming, before anyone has to make a paying decision.
This sits alongside Anthropic's Claude for Teachers, launched in mid-July, which took a different approach. Rather than a blanket free consumer tier, it gave educators tools built around Claude Skills and connectors. Where Google is optimising for volume and habit formation among students, Anthropic's play has been narrower and more structured: equip the people designing the learning experience rather than hand every student a free premium account.
Why this matters even if you don't run a school
The part worth taking from this isn't the education angle. It's the underlying strategy.
Google is willing to give away a genuinely useful product to the group most likely to carry the habit into their working life. That's a land grab, and it's one Australian businesses can learn from without competing in it, because the same logic applies to your own staff.
Most Australian SMBs are not subsidising a year of anything for their team. But the businesses getting real value out of Claude are the ones treating staff AI literacy as a genuine training investment rather than a one-off tool rollout. The habit forms either way. The only question is whether it forms deliberately or by accident.
What deliberate looks like in practice
There's a version of AI rollout that most businesses default to: buy licences, send an announcement email, and hope. Six weeks later usage has settled into two power users, a handful of people using it as a slightly better search box, and everyone else quietly not bothering.
The alternative isn't complicated, it's just intentional:
A scoped Claude Skill for onboarding, so new starters learn your actual processes rather than generic best practice.
A structured internal training pass on how to use Claude Code or Claude Cowork for a specific role, rather than a generic overview that leaves everyone to work out their own use case.
A short compliance or SOP-onboarding build using Claude Skills, giving staff something concrete to practise against instead of a demo they watched once.
A named person responsible for collecting what's working, so good patterns spread instead of staying with whoever discovered them.
Together those do for a workforce roughly what Google's free year is trying to do for students: build the habit properly and early, before people default to using AI badly or not at all.
The cost of letting habits form by accident
Bad AI habits are harder to unwind than no habits at all. A team that has spent six months pasting client information into a general chat window has both a training problem and a Privacy Act problem, and the second one is considerably more expensive to sort out than the first would have been.
The same applies to quality. Staff who learned to accept first drafts uncritically produce work that looks fine and occasionally isn't, and that pattern is hard to spot from above until something goes out the door wrong. Teaching people what to check, and when to push back, is a skill that has to be taught rather than absorbed.
What this costs an Australian business
A structured internal AI training pass for a small team, covering role-specific workflows, a practical governance boundary, and a set of worked examples using your own material, typically runs A$4,000 to A$8,000 depending on team size and how many distinct roles need covering.
Set against the licence spend most businesses are already committing to, that's a modest addition, and it's usually the difference between licences that get used and licences that get renewed out of inertia. A Sydney firm paying for twenty seats and getting genuine use from four is not saving money by skipping the training.
The Australian angle
There's no equivalent free-year play available to most Australian businesses, and there doesn't need to be. What matters is deliberate investment in getting staff comfortable and skilled with the tool you've actually chosen.
That's a smaller, more targeted version of what Google is doing at scale with students. Treat AI literacy as something you build on purpose, tied to real work your team already does, rather than something you assume will happen once the licences are switched on.
It's also worth noticing which strategy Anthropic picked. Equipping the people who design the learning experience, rather than blanketing the end users, is the same instinct that works inside a business: train the people who set how work gets done, and the practice spreads from there with a lot less effort than a company-wide announcement ever generates.
If your team has Claude licences and you're not sure they're being used well, that's usually a training gap rather than a tooling one. Book a session and we'll scope what a role-specific training pass would cover.
