The threat of quantum computers to government systems, how AI is upending math and the ongoing Pentagon-Anthropic saga. All that and more in this week’s Prototype. To get it in your inbox, sign up here .

If, like me, you came of age in the 1990s, you probably remember talk about the Y2K bug–warnings about the sheer chaos of what might happen when software suddenly read the new year as 1900 instead of 2000, wreaking havoc on things like payment systems. Then January 1, 2000 rolled around and… basically nothing happened. Not because the warnings were wrong, but because thousands of people and billions of dollars were deployed to fix the bug before it became an issue.

Now the industry is facing a similar threat: Q-Day–the day that a scalable quantum computer will be capable of breaking the most commonly deployed encryption schemes on the internet. Unlike Y2K, there’s no set date for when this will occur, though it’s widely expected to happen in the 2030s as quantum computers steadily improve .

And like Y2K, we know how to fix the problem. It’s just tedious. There are encryption algorithms out there that are, in theory at least, impossible for quantum computers to crack. But a new GAO report published last week found that federal government systems are not only not ready, they’ve barely even started to address the issue at all. “Federal agencies recognize the quantum computing threat and have taken some actions to partially address it,” the authors wrote. “However, agencies have not fully carried out initial steps to prepare for this threat.”

Why do agencies need to start acting now? Well for one, because while the threat is probably more than a decade away, there’s no guarantee of that–it could happen sooner. Secondly, because federal systems store important data for years and decades, hackers could theoretically steal encrypted data now with the intent of decrypting it with quantum systems when feasible (referred to as “Harvest Now, Decrypt Later”).

It’s not just federal agencies at risk, either. Google is currently aiming to have its systems transferred over to post-quantum cryptography by 2029 , but it’s ahead of the curve. According to a recent report by cybersecurity firm Wiz, 81% of cloud services don’t even allow the option to configure systems to post-quantum algorithms, and less than 1% of companies are currently using post-quantum algorithms for digital signatures that verify users and systems are who they say they are. “Everywhere we look there is still a lot of work to be done,” report author Scott Piper wrote.

Got questions? Comments? Story tips? Book suggestions? Email me at: theprototype@forbes.com

Last week, OpenAI released more than 700 math preprint papers online, purporting to solve a number of long-standing math problems. These were far from formal proofs–less than half had been validated with Lean, the field’s industry standard software. At least three have already been retracted. Questions have also been raised about whether some of these papers were, in part, based on data mathematicians had previously uploaded into ChatGPT.

That said, many of the proofs definitely seem to provide insight into long-held problems. Though it may take some time to figure out–the papers were not presented as formal proofs and in many cases the AI models even appear to have developed their own jargon for handling the problems. “We’re used to landmark results being accompanied by new insights and theories,” MIT professor Ankur Moitra told me via email. “But these feel different. The claims are spectacular, the papers are incomprehensible and there’s no punchline in sight that explains why the mysteries suddenly unraveled.”

This is hardly the first time computers have upended mathematics. In fact, they’ve even changed the word “computer”–which used to refer to human mathematicians who laboriously did complex calculations by hand. There have been a lot of reactions from the mathematician community to this, you can read some of them here , and it will be interesting to see how this shakes out over the coming months and years.

“I can say that I’m impressed by the volume and the importance of the problems solved,” MIT professor (and Forbes 30 Under 30 alumnus ) John Urschel told me. “I’m personally making my way through the 9/4 matrix multiplication paper, though I have many on my reading list. It’s obvious that how we do and evaluate mathematics is going to have to change. What I think is constant is the fundamental idea of understanding that underpins our profession. And I’m excited to learn from these breakthroughs."

Anthropic v The Pentagon, Week 32: The Department of Defense first labeled Anthropic a supply chain risk way back in February, and I’ve been charting the ups and downs since. But it looks like the Pentagon has finally stopped using the AI company’s tools this month, according to the BBC , though it was apparently still using in October.

SpaceX Mobile: SpaceX announced that it has acquired nationwide, low-band wireless spectrum , pending FCC approval. This would make it easier for its satellites to deliver data to cell phones indoors, and most phones are equipped to receive at this frequency. The FCC has also approved the company’s plans to put 15,000 Starlink satellites in orbit to support the effort. But competition is starting to nip at its heels: Amazon has built its 1,000th internet satellite , and it plans to launch its service at the end of the year.

Nobel Prizes: The scientific Nobels were awarded last week. Karl Deisseroth, Peter Hegemann and Georg Nagel won the Medicine prize for their work in optogenetics–making it possible to make light-controlled “switches” for nerve cells–which has made it possible to map neural circuits. Henri Kagan and Kenso Soai won the Chemistry prize for figuring out how to develop chemical reactions for homochiral molecules–ones that all have the same basic structure–in cases where a molecule has the potential to have one of two mirror image forms. That’s important for things like pharmaceuticals, where one version of a chemical can be a useful drug but the other isn’t. And finally, Francis Halzen got the Physics honors for his insane/genius idea to turn Antarctic ice into a giant neutrino detector, which has led to the discovery of multiple types of the elusive particles and expanded our understanding of the universe.

Cutting Astral Red Tape: The FAA is proposing five new rules to streamline the regulatory process for commercial space launches, including electronic submissions and omission of duplicative approvals. There can be a lot of bureaucratic back and forth for launches, especially for companies doing something for the first time, so some streamlining of the process will probably make life easier for space entrepreneurs–depending on what the final rules end up looking like.

AI Financing: Following up on some of last week’s discussion on AI debt, my colleague Brandon Kochkodin has an interesting piece about an AI-powered credit rating firm, AIR Platforms, that grades OpenAI’s debt at junk bond status and Anthropic’s as merely investment grade. Of course, most investors are going to look for Moody’s or S&P ratings–and AIR doesn’t have access to OpenAI’s or Anthropic’s financials–but it’s food for thought. Meanwhile, Apollo’s chief economist Torsten Slok noted last week that despite the billions being spent on AI, there’s no signs of increased productivity in the economy–which definitely raises more ROI questions.

What’s Entertaining Me This Week

A little less than a year ago, I said I was looking forward to Three Hitmen and a Baby , the latest in Rob Hart’s Assassins Anonymous series about a group of former violent secret agents, mercenaries and assassins who’ve joined a self-help group to stop themselves from killing again. Well, now I’ve read it, and it did not disappoint. Once again the past has come back to haunt the main characters, and once again Hart does a fine job of striking a balance between writing a fun thriller while deconstructing the things that we, the audience, enjoy about them, forcing us to confront exactly what kind of violence we’re asking from our fictional characters, and what kinds of real world impact it would have on them. Stay tuned next year for my take on book four of the series, City of Killers , which comes out next summer.