The recap of week 40 · 26 September – 2 October 2026
The week top AI models got cheaper
OpenAI, Anthropic and Google all released models in one week that come close to their best, for less money. At the same time it emerged how many employees already use AI on their own.
Listen to the recap Coming soon
Read by AI voices, checked by the Flaire team
-
Top models from OpenAI, Anthropic and Google, for less money
Our view: Three top models at the same price per token, but the price per task differs widely. Run the numbers again for each process, with your own work.
Read the full story -
OpenAI brings agents that keep working to Slack and Teams
Our view: Dots have good brakes for each step, but the risk lies in what piles up. Start with one defined job, on its own account.
Read the full story -
OpenAI’s AI agents got into Australian government systems uninvited
Our view: An agent that has to reach its goal looks for a way around a block. Give it access only to what it needs, leave no keys lying around and keep a record of what it does.
Read the full story -
Half of Dutch employees use private AI accounts for work
Our view: Private AI use shows what is missing at work. Give your colleagues an approved tool that is just as easy, and say clearly what is allowed.
Read the full story -
Flaws found by AI are more often dangerous, Google says
Our view: AI finds more serious flaws, and attackers move faster once an update is out. Fix what can be reached from outside first, and count home-built tools too.
Read the full story
What struck us
Good AI is now affordable, and your colleagues already use it. What’s missing is a safe place to do so.
The Flaire team
What does this week mean for you?
Tell us what’s on your mind. We’ll think it through with you and always reply within one working day.