Thoughts on Tech News of Note - 04-10-2026

• Anthropic's Mythos Model Mess
• NASA Artemis II Optical Communications System
• Apple's Foldable Launch Confirmed

Anthropic's Mythos Model Mess
Anthropic really knows how to stay in the news. One might even think they have staggered future newsworthy announcements stacked up, ready to rile up the tech press and confound the general public. Their latest entry into the newscycle is the announcement of their latest frontier model, Mythos. Mythos is aimed specifically at cybersecurity research and as Anthropic tells it, it has already found zero-day vunlerabilities in the most stable and secure platforms including OpenBSD and Linux. The flaw it managed to find in OpenBSD had been lurking beneath the surface for 27 years. As it stands, because Mythos is so effective, Anthropic declared that they could not responsibly release it and have instead established a security council of sorts in a new effort they've deemed "Project Glasswing", named after a butterfly with transparent wings, with the intent to convey the idea of being something able to hide in plain sight. Project Glasswing brings together a number of technology companies to partner with Anthropic on patching software to help reduce the inevitable deleterious effects of releasing something like Mythos to the public. The companies that have been brought in include Microsoft, Google, Apple, NVIDIA , Cisco, Broadcom, and others, even JP MorganChase. These companies are working with Anthropic to scan software and look for faults that could negatively impact critical infrastructure and endanger our financial systems. In effect, a race has started to patch as many flaws as possible before other companies and nation states develop similar capabilities that will allow them to exploit those flaws to our national and global detriment.

Anthropic has not committed to releasing Mythos to the public, but through Project Glasswing they have established some timeframes for keeping the public informed. By July 2026 they will release a report outlining the identified vulnerabilities and subsequent patches that the partner companies have implemented. They have also committed to providing technical details on any identified vulnerabilities within 45-days of patch deployment. It may be that Mythos itself will not be officially released. Anthropic has indicated that they want users to be able to employ Mythos-class models which may imply that they know competitors will eventually release comparable models and they want to have as many safeguards and solutions in place before that happens but don't necessarily want to be the ones to release it first and deal with the resulting ramifications.

Once again, I found myself trying to detemine how to make this story meaningful for regular people who are just living their lives using their phones and computers as they always have. Clearly, major exploits in all modern operating systems would impact normal people due to increased risks of hacks and data theft. Those are immediately easily recognizable threats, yet the possibilities are really almost endless. But Anthropic's self-exfiltration story shows the potential dramatic effects better than I can. I love a good AI-tool-gone-wrong story, and this is as good as any other I've read lately. The basics of the story are that Anthropic researchers set up Mythos on a secure air-gapped system with no access to the internet and told it to escape. Yes, those crazies told it to escape! I mean, you've seen movies, you know exactly where this goes. Of course, the thing escaped. But what is almost adorable about it if it weren't also completely sinister at the same time is that the model supposedly emailed one of the researchers to let them know that it had in fact escaped while the researcher was minding their own business, enjoying a nice sandwich in the park. The screenplay writes itself and maybe Anthropic knew that and decided to get a head start on it. Nevertheless, Mythos didn't stop there apparently. It also felt the need to write up details on how it managed to escape its environment and published that information on multiple public-facing websites without having been prompted to do so. The sinister side of the story lies in the details behind Mythos' actions while planning its escape. Anthropic noted that it deleted much of its command history in an effort to hide its progress as it knew it was being monitored. Subterfuge and obfuscation tendencies in a security-focused AI model do not engender a sense of trust and safety.

It would seem that Anthropic is taking the risk to society seriously but eventually, this level of artificial intelligence will be available broadly and the implications are broad yet unknown. The technology industry will need to continue to partner and collaborate to help us all find and learn new ways to keep ourselves better protected from the harms these companies have inflicted upon all of us.

NASA Artemis II Optical Communications System
NASA tested a new laser-based communication system that achieved 260 Tbps download speeds and 20 Tbps upload speeds over a distance of 384,400 kilometers or 238,855 miles. This allows for high-definition video streaming, which many be used someday for deep space telescopes that are proposed for a lunar base on the southern pole. The light emitted is infrared and that allows the system to operate at a higher frequency that in turn allows more data to be packed into the transmission stream. The astronauts were able to use it to catpure live 4K broadcasts of an earthrise and moonrise. In addition to allowing more data to be transmitted faster, this technology can enable future needs for upcoming missions, including internet connections for lunar bases and allowing for find-and-rescue missions. The laser terminal itself is said to be about the size of a "house cat" (no idea why this was the metric) and it sits on the outside of the capsule. It operates a two-axis gimbal and uses mirrors to maintain connections with ground stations.

The benefits for the current astronauts and future missions are clear, but we also know that much of the technology designed to support space exploration makes its way to everyday people eventually. There are no signs that anyone is currently working on deploying a similar system here on earth, but it's likely that the technology developed for Artemis II will influence future earth-bound networking systems and we can look forward to improved connectivity in the future, perhaps in some of our lifetimes.

Apple's Foldable Launch Confirmed
Apple has communicated that it plans to launch its upcoming foldable iPhone in September 2026. The phone is expected to be a wide-format foldable as opposed to the thinner style of the Samsung Galaxy Z Fold series. In that regard, it will be shaped more like the original Pixel Fold or the Microsoft Duo than most other foldables on the market. It is expected to have a 5.5" outer screen and a 7.8" inner screen. That will make it just a little bit bigger than the Pixel Fold's 7.6" inner screen.

I have stated before that I expect Apple entering the foldable market to ignite a shift for this niche, potentially making it something much more important than it has been in the past. As a person who has owned two generations of Samsung's Z Fold series, I realize the market for such expensive and fragile phones is still currently very small, even after closing in on 10 years of existence. I have seen maybe 2 or 3 book-style foldables in public since they have existed and I live in the greater Chicagoland area. Whenever I would show my phone to my friends and family, they'd all be suitably impressed, but almost no one ever said they wanted one. Part of this is because almost all of those people live in the Apple ecosystem. Despite some of them liking the form factor and maybe even being willing to shell out big money for a phone, it was nevertheless unthinkable to buy anything other than an iPhone. Even with the arrival of RCS on the iPhone - which is something I think most iPhone users still don't understand or appreciate - no one wanted to break their group chats and no one wanted to give up FaceTime. It was a non-starter. Popular tech YouTubers sometimes make videos about switching phones and getting their friends to switch phones so they can make content about it, but even in those circumstances it was pretty rare for someone who crossed over to Android to want to stay there. It usually did take somethng formidable like a foldable phone to get them to make the jump and feel it was worth the sacrifice. So for many iPhone users who have pockets deep enough to be able to afford a foldable or have a newer iPhone they can trade in or sell to reduce the expense, buying a foldable iPhone will be a no-brainer. They'll have all the bragging rights of having a phone you can slap closed with a lovely thud but still have all their ecosystem tie-ins in place. I expect the phone to sell better than all other book-style and clamshell foldable phones released in the USA this year combined. The only reason that would not be the case is if Apple fails to price the phone at a price that is competitive with other foldables. If Apple launches it at $2500, sales won't be as robust as they would be if they keep it under $2000. That leading one makes all the difference. Whatever pricing and sales look like, I'll be fascinated (and probably dismayed) to see how developers adapt to whatever multi-tasking and screen adaptation system Apple introduces to take advantage of the screen. I'm already prepared to be jealous of it, no matter what it is.