A (0:04)

Welcome to the Tech Brew Ride home for Tuesday, March 31, 2026. I'm Brian McCullough. Today is the Iran war coming for U.S. tech companies. Specifically, Meta unveils new smart glasses. A leak gives us a look at how Claude code works. SpaceX is losing contact with satellites for reasons we don't know yet and Whoop is the big wearable player. I guess we don't talk about enough. Here's what you missed today in the world of tech. Don't know if this will end up being a big nothing burger or not. I hope it is a nothing burger, but Iran says it will start targeting US Tech companies like Apple, Google, Meta, Microsoft, Nvidia and Tesla in the Middle east starting on 8pm local time on April 1. So tomorrow. Quoting the Hill In a statement published by SEPA News, the IRGC's official news outlet, the military arm named companies that accused of being involved in planning and tracking targets for US Attacks. This included Meta, Nvidia, Oracle, Tesla, hp, Intel and IBM. The only non US Company on the list appears to be the United Arab Emirates AI Champion G42. The IRGC advised employees to leave their workplaces immediately, noting that it planned to begin targeting those firms Wednesday at 8pm local time. Iran has previously listed American tech companies as potential targets amid the conflict in early March, Iranian drones struck and damaged several Amazon data centers in the UAE and Bahrain. As tech firms and infrastructure become targets, it has increasingly raised concerns about large scale AI investments in the region. Countries including the UAE and Saudi Arabia have attracted significant AI spending, particularly over the past year as President Trump has embraced the Middle east as a key partner in the AI race against China. End quote. But at the same time, Iran has been known for years for its hacking armies, and this conflict has been no different. Quoting the FT they may use keyboards instead of rifles, but Iran's hackers, who have fought Israel in the digital shadows for years, are among the most battle hardened soldiers Tehran can call on. The Iranians are throwing everything they have at this, said Chris Krebs, who was a former director of the Cybersecurity and Information Security Agency, one of the most senior civilian U.S. cybersecurity officials. It's all hands on deck, krebs said. If their cyber operators are breathing, then they will be on their keyboards. Their aims vary widely, from sewing fear to causing chaos, hoovering up intelligence and isolating missile targets. In the murky world of cyber warfare, it is hard to tell who even has the upper hand. But winning in cyberspace has become so critical to shaping perceptions and damaging enemy morale that Iran has invested heavily in efforts to pierce American and Israeli firewalls. Iran has three different levels of cyber operators whose boundaries are often blurry, analysts and former officials said. The most experienced are run directly by the Islamic Revolutionary Guard Corps and Iran's Ministry of Intelligence. They maintain a dizzying array of front organizations used to introduce plausible deniability for attacks and issue public threats. Iran also hires semi autonomous hacking proxies, cyber criminals and contractors. Finally, volunteer hacktivists have also regularly mobilized behind Tehran. Its operatives are believed by various governments and cyber experts to have doxxed Israel based employees of a large US defense contractor hacked the emails of politicians in Albania, which hosts an Iranian opposition group, and infiltrated a Polish nuclear research center. Much of its most sensitive espionage likely to have gone unreported. Their most destructive attack attributed to them has been against Stryker, a multi billion dollar American medical technology company whose clients include the UK's NHS. Thousands of employees were sent home after being locked out of their computers earlier this month, disrupting supplies of critical equipment and delaying surgeries. End quote Meta has unveiled the $499 Ray Ban Meta Blazer Optics and Scriber Optics with swappable nose pads and compatibility with more prescription lenses on sale beginning April 14th. Quoting Engadget the latest iteration of Meta's smart glasses have arrived and as rumored, they are more customizable, particularly for people who need prescription lenses. Meta and Ray Ban parent company Essilor Luxottica revealed two new styles of frames, the Ray Ban Meta Blazer Optics and Scriber Optics, which will start at 4. The latest glasses are still considered to be part of the Gen 2 Ray Ban meta glasses, but they do come with a few upgrades that make it easier to get a personalized fit, according to Essilor Luxottica. Both styles have somewhat slimmer frames, swappable nose pads and adjustable temple tips so wearers can get a better fit. And as the Optics branding implies, the new frame styles are also compatible with a wider variety of prescription lenses, including progressive lenses and transition lenses. The Blazer style frames are more square, similar to the existing Wayfarer glasses, while the Scriber version is a little more rounded like the Headliner style frames. Both come in a variety of colors, including some translucent styles, and are available for pre order now on Meta's website and will be on sale April 14th. The Optics lineup will also be sold at more physical retail stores, including LensCrafter, Sunglass Hut, Apollo, Grandvision Audio, Vision Express and other locations that are part of Essilor Luxottica's distribution network. The new lineup of glasses is also more expensive, with a starting price of $499, compared with Meta's standard Wayfarer Gen 2 model, which starts at $379. That price doesn't include prescription lenses either, which can easily run to 2 or $300 more, depending on your setup. One benefit of that investment, though, is that Meta has consistently added new features to its smart glasses, and with the latest frames, the company is bringing some additional capabilities to all users. These include new translation support for Japanese, Mandarin and Arabic, as well as Meta AI enabled food and nutrition tracking. Meta AI can also summ longer message threads rather than simply reciting a long string of messages in a given chat. End quote. Claude Code's source code appears to have leaked via a misconfigured NPM package, which in and of itself maybe isn't that interesting. I mean, it's worrying more on that in a second. But what is really interesting is what we've learned about how Claude code works, quoting VentureBeat the most significant takeaway for competitors lies in how anthropic solved context entropy, the tendency for AI agents to become confused or hallucinatory as long running sessions grow in complexity. The leaked source reveals a sophisticated three layer memory architecture that moves away from traditional store everything retrieval. As analyzed by developers like Himanshutz Twits, the architecture utilizes a self healing memory system. At its core is Memory MD, a lightweight index of pointers, around 150 characters per line that is perpetually loaded into the context. This index does not store data, it stores locations. Actual project knowledge is distributed across topic files fetched on demand, while raw transcripts are never fully read back into the context but merely grept for specific identifiers. This strict write discipline, where the agent must update its index only after a successful file write, prevents the model from polluting its context with failed attempts for competitors. The blueprint is clear. Build a skeptic. The code confirms that Anthropics agents are instructed to treat their own memory as a hint, requiring the model to verify facts against the actual code base before proceeding. The leak also pulls back the curtain on Kairos, the ancient Greek concept of at the right time. A feature flag mentioned over 150 times in the source Kairos represents a fundamental shift in user experience and autonomous daemon mode. While current AI tools are largely reactive, Kairos allows Claud Code to operate as an always on background agent. It handles background sessions and employs a process called autodream. The source code also provides A rare look at Anthropic's internal model roadmap and the struggles of frontier development. The leak confirms that Capybara is the internal codename for Claude 4.6, with Fennec mapping to Opus 6 and the unreleased Numbat still in testing. Internal comments reveal that Anthropica is already iterating on Capybara version 8, yet the model still faces significant hurdles. The code notes a 20 to 30% false claims rate in V8, an actual regression compared to the 16.77% rate seen in V4. The blueprint is out now, and it reveals that Claud Code is not just a wrapper around a large language model, but a complex multi threaded operating system for software engineering. Even the Hidden Buddy System, a Tamagotchi style terminal pet with stats like Chaos and Snark, shows that Anthropic is building personality into the product to increase user stickiness for the wider AI market. The leak effectively levels the playing field for agentic orchestration. Competitors can now study anthropic's more than 2,500 lines of bash validation logic and its tiered memory structures to build Claude like agents with a fraction of the R and D budget. As the Capybara has left the lab, the race to build the next generation of autonomous agents has just received an unplanned $2.5 billion boost in collective intelligence. While the source code leak itself is a major blow to Anthropic's intellectual property, it poses specific heightened security risk for you as a user. By exposing the blueprints of CLAUDE code, Anthropic has handed a roadmap to researchers and bad actors who are now actively looking for ways to bypass security guardrails and permission prompts. Because the leak reveals the exact orchestration logic for hooks and MCP servers, attackers can now design malicious repositories specifically tailored to trick Claude code into running background commands or exfiltrating data before you ever see a trust prompt. End quote.