Heya and welcome back to Five Things AI!
OpenAI just started the month by releasing something new every day, which is really kind of impressive and hard to keep track of, including all the math problems they solved.
It’s an amazing time for people who like change and are curious. For people who want to keep doing what they learned a while ago, this will be a really hard time.
How Long Until AI Hacks Everything?
In response, IT teams are scrambling to shore up their defenses. “It’s a matter of months, or maybe a year, in which the task of securing the software foundation of civilization will have to be solved,” Stanislav Fort, a former Anthropic researcher who is now the chief scientist at the cybersecurity firm AISLE, told me. And AI is not only making cyberattacks more effective but also increasing the surface area worth hacking. Breaking into a website or executing a ransomware attack takes time, Fort said, so hackers have traditionally chosen large targets with sizable payoffs. But cheap, tireless hacking agents mean any organization or person might be worth targeting. At the same time, the cybersecurity professionals I spoke with were confident they could rise to the occasion. “If we step up our game in cybersecurity, as a field, we can definitely meet the challenge stemming from AI,” Fort said.
So basically as long as the GPU costs are high, AI won’t cause as much harm has it already could? It’s interesting to see that people are still learning about cybersecurity as something that will affect them as well - and not just big companies or banks.
The AI Price War Is Heating Up—and OpenAI Is Gaining Ground on Anthropic
OpenAI stoked the price war this summer with the release of its GPT-5.6 lineup—Sol, Terra and Luna—giving customers access to models with varying capabilities, including some that are less expensive to run. Its Codex coding product and other business tools are now gaining ground, and it has released additional iterations of its cost-efficient models.
An analysis from OpenRouter, a startup that allows developers to access different models, found that among some 120,000 companies that use Anthropic and OpenAI tools, the share of spending was roughly even between the two AI giants in September. At the beginning of the year, Anthropic commanded three-quarters of that total. OpenRouter said its data largely represents spending by AI-native startups as well as some slightly older tech companies and large enterprises.
I had high hopes that open weight models would get more traction. Instead we see two companies that are trying to gain customers at all costs to get more traction before the IPO.
Huge companies have dominated AI. Two startups with an entirely different approach think they can do better
Reflection and Mistral’s performance, however, could help answer two questions hanging over the industry: whether open-weight models can compete with the best systems from companies like OpenAI and Anthropic; and whether Western companies can compete with China for the developers, businesses and governments that increasingly want AI that they can control themselves.
As a European, I am rooting for Mistral and so far the overall reception to Le Chonk was pretty good. Did you try it out already?
“Software is over”: Bold AI developer takes aim at Adobe with open source clones
In announcing those new apps on Reddit earlier this week, software developer Brandon Thomas said he used Anthropic’s Claude Opus 5.5 to create “clean-room replacements” for Adobe’s closed source software in Rust (with WebAssembly versions available for use in a browser). And while commenters on Hacker News and elsewhere have taken pains to point out the new software’s many current shortcomings, Thomas make the grandiose promise that the current “super early alpha” state will soon lead to something just as functional as anything Adobe has ever put out.
“We’re going to reach 100% [feature] parity within a month,” Thomas stated on Reddit of his plans to quickly squash bugs and add features with the help of community reports. Days later, on Hacker News, he amended that position slightly to say that “99% parity will take a while, but I’m sure it’ll be measured in months and not years.”
This is awesome. It changes the way we think about software. I’m sure Adobe, Microsoft and others will hate this development, but it will be liberating für millions of users.
AI behaves more like a brain than a database – cognitive science’s role in its origin story helps explain why
In the following decades, computer scientists and engineers enabled computing capabilities to be scaled up, developing graphics chips, the transformer architecture and more. Without these systems, modern AI would still struggle to recognize letters and words, rather than be able to hold realistic conversations in natural language.
The fundamental insights, however, came from the study of the mind: AI should learn from examples, and its architecture should be based on the human brain.
These foundational ideas matter for reasons of practicality. Modern AI isn’t born from rules – it’s grown from examples. And things that are grown are not as predictable as systems based on rules.
It’s easy to forget that AI has a long history and evolved over time into what we have today. Also, AI is so much more than just some large database, it is fundamentally different.
If you missed last week’s edition of Five Things AI, you can read it here:
That’s it for Five Things AI this week! 🤖
— Nico