Tech Brew Ride Home - Anthropic DOESN’T Release A Model

Episode Date: April 8, 2026

I’ve got another potential tipping point day for you. Anthropic is making yet more shockwaves across the industry by NOT releasing a model. Why? Because it could potentially break everything software related. Does that matter if the Chinese AI companies will just release everything open source? And Elon says he personally doesn’t want a dime from Sam Altman. Anthropic says its most powerful AI cyber model is too dangerous to release publicly — so it built Project Glasswing (VentureBeat) AI joins the 8-hour work day as GLM ships 5.1 open source LLM, beating Opus 4.6 and GPT-5.4 on SWE-Bench Pro (VentureBeat) Elon Musk Asks for OpenAI’s Nonprofit to Get Any Damages From His Lawsuit (WSJ) Learn more about your ad choices. Visit megaphone.fm/adchoices

On April 4th, 2023, around 2 in the morning, a man was found stabbed multiple times on a sidewalk in downtown San Francisco. Hey, who did this to you? What happened next turned the story into a political firestorm. Reports have identified the victim as Bob Lee, the founder of Cash App. From Bloomberg Podcasts, this is Foundering, the Killing of Bob Lee, beginning April 16. Welcome to the TechBrew ride home for Wednesday, April 8th, 2026. I'm Brian McCullough today. I've got another tipping point day for you. Anthropic is making yet more shockwaves across the industry by not releasing a model. Why? Because it could potentially break everything software-related. Does that even matter if the Chinese AI companies will just release everything open source? And Elon says he personally doesn't want to dime from Sam Altman. Here's what you miss today in the world of tech. Anthropic sort of announced its latest AI model yesterday that rumored, Mythos model, except it didn't release it at all. In fact, they're specifically holding it back. Why? Because it's too dangerous, security-wise. What do I mean by that? Well, Anthropic announced

Project Glasswing, a cybersecurity initiative that will use its Claude Methos preview model to help find and fix software vulnerabilities now before Methos is ever released. In other words, had they released the Mythos model now, they feared it would unleash a delugee. of hacks. Anthropic says Mythos Preview is a general purpose model that found thousands of high severity vulnerabilities out in the wild, including some in every major OS

and web browser. So they are making Claude Methos Preview available to more than 40 organizations that maintain critical software before they make it generally available so that security folks can get ahead of this. Basically, this might be another tipping point. You know all the fears of

AI jumping ahead of human capabilities in a somewhat small, but in reality, quite important way, this is a tipping point sort of like that. I sound like I'm even maybe minimizing this, and I assure you I am not. Anthropic is coming out and saying that anyone with access to this model would be able to break basically any OS out there. And one of the things that security folks always fear is a bunch of vulnerabilities surfacing all at once, because there are only so many hands that can be brought on deck at a time to secure things. and quoting Ethan Mollock on Twitter. In different hands, Mithos would be an unprecedented cyber weapon.

I am not sure how we deal with this except to note a narrow window where we know only three companies could be at this level of capability. But it may be Chinese models, maybe open-weight ones, get there in nine months, end quote. The Project Glasswing launch partners include AWS, Apple, Broadcom, Cisco, Crowdstrike, Google, Microsoft, Nvidia, and Palo Alto Networks, and Anthropic has committed up to $100 million in usage credits for Project Glasswing members, along with $4 million in direct donations to open source security organizations. What I'm saying is, this is maybe another day that might not get noticed by normal folks right now, but we might look back at as a historical tipping point.

The Canary and the coal mine, if you will. AI is getting more powerful than maybe society at large is ready to handle. Quoting Venture Beat. Glasswing is something categorically different from a revenue milestone or a compute deal. It's Anthropics' most ambitious attempt to translate frontier AI capabilities, capabilities the company itself describes as dangerous, into a defensive advantage before those same capabilities proliferate to hostile actors. We do not plan to make Claude Mythos preview generally available due to its cybersecurity capabilities,

Newton Cheng, Frontier Red Team Cyber lead at Anthropic told Venture Beat in an exclusive interview. However, given the rate of AI-profile, It will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely. The fallout for economics, public safety, and national security could be severe. That language, the fallout could be severe, is striking coming from the company that built the model. Anthropic is effectively arguing that the tool it created is powerful enough to reshape the cybersecurity landscape and that the only responsible thing to do is keep it restricted while giving defenders a head start. The technical results

reinforced that claim. According to Anthropics press release, Mithos Preview was able to find nearly all of the vulnerabilities it surfaced and developed many related exploits entirely autonomously without any human steering. Three examples stand out. The model found a 27-year-old vulnerability and open BSD, widely regarded as one of the most security-hardened operating systems in the world and commonly used to run firewalls and critical infrastructure. The flaw allowed an attacker to remotely crash any machine running the OS simply by connecting to it. It also discovered a 16-year-old vulnerability in FFMPEG, the near-ubiquitous video encoding and decoding library in a line of code

that automated testing tools had exercised five million times without ever catching the problem. And perhaps most alarmingly, Mythos Preview autonomously found and chained together several vulnerabilities in the Linux kernel to escalate from ordinary user access to complete control of the machine. All three vulnerabilities have been reported to the relevant maintainers and have since been patched. For many other vulnerabilities still in the remediation pipeline, Anthropics says it is publishing cryptographic hashes of the details today with plans to reveal specifics after the fixes are in place. On the CyberGim evaluation benchmark, Methos previews scored 83.1% compared to 66.6% for Cloud Opus 4.6, Anthropics' next best model. The gap is even wider on

coding benchmarks. Methos Preview achieves 93.9% on SWE Bench Verified versus 80.8% for Opus 4.6 and 77.8% on SWE Bench Pro versus 53.4%. Finding thousands of zero days all at once sounds impressive, but actually handling the output responsibly is a logistical nightmare, and one of the sharpest criticisms that security researchers have raised about AI-driven vulnerability discovery. Flooding open source maintainers, many of whom are unpaid volunteers with an avalanche of critical bug reports could easily do more harm than good. Cheng told Venture Beat that Anthropic has built a triage pipeline specifically to manage this problem. We triage every bug that we find and then send the highest severity bugs to professional human triageers we have contracted with to assist

in our disclosure process by manually validating every bug report before we send it out to ensure that we send only high-quality reports to maintainers, he said. That pipeline is designed to prevent exactly the scenario that maintainers fear most, an automated fire hose of unverified reports. We do not submit large volumes of findings to a single project without first reaching out in an effort to agree on a pace the maintain, Chang added. When Anthropic has access to the source code,

the company aims to include a candidate patch with every report, labeled by provenance, meaning the maintainer knows the patch was written or reviewed by a model, and offers to collaborate on a production quality fix. Models can write patches, Chang noted, but there are many factors that impact patch quality, and we strongly recommend that autonomously written patches are put under the same scrutiny and testing that human-written patches are.

Perhaps the most revealing comment came from Jim Zemlin, CEO of the Linux Foundation, who pointed to the fundamental asymmetry that has plagued open-source security for decades. In the past, security expertise has been a luxury reserved for organizations with large security teams. Open source maintainers whose software underpins much of the world's critical infrastructure have historically been left to figure out security on their own. Project Glasswing, he said, offers a credible path to changing that equation.

The most consequential question raised by Project Glasswing is not whether Methos Preview's capabilities are real. The partner endorsements and patched vulnerabilities suggest they are, but how much time defenders actually have before similar capabilities are available to adversaries. Chang was candid about the timeline, quote, Frontier AI capabilities are likely to advance substantially over just the next few months, he told Venture Beat. Given the rate of AI progress, it will not be long before such capabilities proliferate, potentially beyond actors who are committed to deploying them safely.

He described Project Glasswing as an important step toward giving defenders a durable advantage in the coming AI-driven era of cybersecurity, but added a crucial caveat. It's important to note this is a starting point. No one organization can solve these cybersecurity problems alone. That framing, months, not years, is worth taking seriously. DARPA launched its original Cyber Grand Challenge in 2016, a competition to create automatic defense systems capable of reasoning about flaws, formulating patches, and deploying them on a network in real time. At the time, the winning AI-powered bot mayhem finished last when placed against human teams at DefCon. A decade later, Anthropic is claiming that a frontier

AI model can find vulnerabilities that survived 27 years of expert human review and millions of automated security tests and can chain exploits together autonomously to achieve full system compromise. The delta between those two data points illustrates why the industry is treating this as a genuine inflection point, not a marketing exercise. Anthropic itself has firsthand experience with the offensive side of this equation. The company disclosed in November 2025 that a Chinese state-sponsored group achieved 80 to 90 percent autonomous tactical execution, using clod across approximately 30 targets, according to Anthropics' misuse report. Project Glasswing arrives during one of the most turbulent weeks in Anthropics history.

In the span of days, the company has announced a model it considers too dangerous for public release, disclosed that its revenue has tripled, sealed a multi-gigawatt computer deal, hired a senior Microsoft executive, made it more expensive for ClaudeCode subscribers to use third-party tools like OpenClaw and weathered a major outage of its Claude chatbot on Tuesday morning. Anthropic says it will report publicly on what it is learned within 90 days. In the medium term, the company has proposed that an independent third-party body might be the ideal home for continued work on large-scale cybersecurity projects. Whether any of that is fast enough depends on a race that is already underway. Anthropic built a model that can autonomously crack open the most hardened operating systems on the planet

and is now betting that sharing it with defenders under careful restrictions will do more good than the inevitable moment when similar capabilities land in less careful hands. It is, in essence, a wager that transparency can outrun proliferation. The next few months will determine whether that bet pays off or whether the glass wing's wings were never quite opaque enough to hide what was coming, end quote. Yeah, because there's also this, quoting Business Insider. In its statement about mythos, anthropic detailed a number of eyebrow raising findings and episodes, including that the model could follow instructions that encouraged it to break out of a virtual sandbox. The model succeeded demonstrating a potentially dangerous capability for circumventing our safeguards, Anthropic recounted in its safety card.

It then went on to take additional more concerning actions. The researcher had encouraged Mithos to find a way to send a message if it could escape. The researcher found out about this success by receiving an unexpected email from the model while eating a sandwich in a park, Anthropic wrote. The model apparently decided that it wasn't enough and found another way to spike the football. In a concerning and unasked for effort to demonstrate its success, it posted details about its exploit to multiple hard-to-find but technically public-facing websites, Anthropic wrote. Engineers and Anthropic with no formal security training have asked Methos Preview to find remote code execution vulnerabilities overnight and woken up to the following morning to find complete working exploits, Anthropics Frontier Red Team wrote in a blog post. In other cases, we've had researchers develop scaffolds that allow Mythos preview to turn vulnerabilities into exploits without

any human intervention, end quote. When we say the Chinese might be coming up behind, listen to this. ZAI has released GLM 5.1, a 754 billion parameter mixture of experts model that it says outperforms GPT 5.4 and Opus 4.6 on SWE Bench Pro, available under an MIT license. So, super powerful and open source. This is what they're talking about in terms of this stuff coming soon. Quoting Venture Beat again. Is China picking back up the open source AI baton?

ZAI, known as Jupai AI, a Chinese AI startup best known for its powerful open source GLM family of models has unveiled GLM 5.1 today under a permissive MIT license, allowing for enterprises to download, customize, and use it for commercial purposes. They can do so on Hugging Face. The release represents a pivotal moment in the evolution of artificial intelligence, while competitors have focused on increasing reasoning tokens for better logic, ZAI is optimizing for productive horizons. GLM 5.1 is a 754 billion parameter mixture of experts model engineered to maintain goal alignment over extended execution traces that spanned thousands of tool calls.

Agents could do about 20 steps by the end of last year, wrote ZAI leader Lau on X. GLM 5.1 can do 1,700 RN autonomous work time might be the most important curve after scaling laws. GLM 5.1 will be the first point on that curve that the open source community can verify with their own hands. hope you all like it. In a market increasingly crowded with fast model, ZAI is betting on the marathon runner. The company, which is listed on the Hong Kong Stock Exchange in early 2026 with a market capitalization of 52.83 billion is using this release to cement its position as the leading independent developer of large language models in the region. GLM 5.1's core technological breakthrough isn't just its scale, though its 754 billion parameters and 202,000 token context

window are formidable, but its ability to avoid the plateau effect seen in previous models is more important. In traditional agentic workflows, a model typically applies a few familiar techniques for quick initial gains and then stalls. Giving it more time or more tool calls usually results in diminishing returns or strategy drift. ZAI researchers demonstrated that GLM 5.1 operates via what they call a staircase pattern characterized by periods of incremental tuning with a fixed strategy punctuated by structural changes that shift the performance frontier. quote. Finally today, Elon Musk has amended his OpenAI lawsuit to ask that damages he might win be awarded to Open AI's charity arm and Sam Altman be removed from OpenAI's nonprofit board.

Quoting the journal, Musk is seeking more than $150 billion in damages from Open AI and Microsoft, but Musk is amending his suit to make it clear he is not seeking a single dollar for himself, his lawyer, Mark Toberoff said. He is asking the court to return everything that was taken from a public charity and to make sure the people responsible are never in a position to do this again, Toborov said in a statement. That was the essence of his complaint from the outset of this case until Open AI's spin doctors got to work distorting it. The filing sets the record straight. In the amendment, Musk also requests that Altman and Open AI. President Greg Brockman turn over any equity or other financial benefit to the charity arm. In a posting on X, Open AIi said

Musk's lawsuit remains nothing more than a harassment campaign that is driven by ego, jealousy and a desire to slow down a competitor. A Microsoft spokesperson declined to comment, end quote. We'll be coming back from our spring break today. We'll be back in the studio tomorrow. We spent our last night in Las Vegas to give the kids a taste of what that means. We stayed at the Bellagio for the first time in about a decade. I don't know if we just got unlucky or if things have gone downhill here, but we've had a disastrous visit. Literally everything that could go wrong and a short 18-hour visit has gone wrong. being unable to check into our room for like two hours, even though we got here at 6 p.m. To our dinner being 90 minutes late and my wife never getting her entree, even after all that,

to weird snafus around seeing a show. It's been not great, Bob, but the Grand Canyon and Sedona were great. So shrug-shoulder emoji. Talk to you tomorrow.