AI Summary
5 min readAnthropic’s Dangerous Model Stays Locked Up—For Now
Anthropic announced a new AI model called Mythos Preview, then immediately refused to release it to the public. The reason: the company says the model is so good at finding software vulnerabilities that giving anyone access to it would amount to handing out an “unprecedented cyber weapon.” Instead, Anthropic launched Project Glasswing, a cybersecurity initiative that shares the model with a restricted set of 40+ organizations maintaining critical software—including AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, Microsoft, NVIDIA, and Palo Alto Networks—so defenders can patch holes before attackers can exploit them. The company has committed up to $100 million in usage credits for Glasswing members and $4 million in direct donations to open source security organizations.
What Mythos Preview Actually Found
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Tech Brew Ride Home
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:40) **Anthropic Withholds Mythos Model Over Cybersecurity Fears** - Anthropic announces Project Glasswing, a cybersecurity initiative that uses its new Mythos Preview model to find and fix software vulnerabilities before the model is released to the public, because they fear it could "break everything software related."
- 2 (02:12) **Why This Is a Tipping Point** - The host frames the announcement as a potential historical tipping point where AI capabilities have outpaced society's readiness.
- 3 (03:48) **Anthropic's Official Stance: "Fallout Could Be Severe"** - Newton Cheng, Frontier Red Team Cyberlead at Anthropic, states the company does not plan to make Mythos Preview generally available due to its cybersecurity capabilities.
- 4 (04:29) **Three Stunning Technical Examples of Mythos Preview's Capability** - The model autonomously found and exploited vulnerabilities that had survived for decades.
- 5 (05:43) **Benchmark Performance: Mythos vs. Opus 4.6** - Mythos Preview scores 83.1% on the Cyber Gym evaluation benchmark vs. 66.6% for Opus 4.6, and 93.9% on SWE-bench verified vs. 80.8%.
- 6 (06:12) **The Logistical Nightmare of Responsible Disclosure** - Flooding unpaid open-source maintainers with critical bug reports could do more harm than good, so Anthropic built a triage pipeline.
- 7 (07:35) **The Fundamental Asymmetry in Open Source Security** - Jim Zemlin, CEO of the Linux Foundation, notes that security expertise has historically been a luxury for organizations with large security teams, leaving open-source maintainers to figure it out on their own.
+ Full timestamped outline available in the app
Show Notes
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
More from this podcast
Tech Brew Ride Home →