← Back to articles

We have a year to fix security everywhere

2026-09-08 | original

read original get YubiKey 5 NFC Security Key → more articles

Why This Matters

The post argues that the release of GLM 5.3-flash — an open-weight, cheap-to-run model near the frontier — collapses the barrier to capable offensive hacking, because abliterated versions with safety refusals stripped out are freely available. If true, the industry's window to find and patch vulnerabilities at scale is short, though the same frontier models can be used defensively. It's a concrete framing of the dual-use AI security debate that matters to anyone running software infrastructure.

Key Takeaways

Open-weight models like GLM 5.3-flash can be downloaded, modified, and run locally for roughly $5–15k in hardware, putting near-frontier capability in anyone's hands.
Groups such as DeAlignAI 'abliterate' safety refusals from released models; the author cites a 0% refusal score on Harmbench-320 for cybercrime and disinformation prompts.
The author's proposed response is defensive: use frontier LLMs to find and fix vulnerabilities industry-wide (Project Glasswing, Daybreak), with deployment — not discovery — as the hard part.

Worth a Look

YubiKey 5 NFC Security Key — If the article's warning about cheap AI-powered attacks lands, hardware-backed authentication is the single most practical personal fix. The YubiKey 5 NFC gives you phishing-resistant logins over USB-A and NFC across accounts that support FIDO2/WebAuthn, so stolen passwords alone won't get anyone in.

See YubiKey 5 NFC Security Key on Amazon → Affiliate link — we may earn a commission on purchases, at no extra cost to you. Product picked by AI based on this article; it is not a tested recommendation.

GLM 5.3-flash released last week, and that means Project Glasswing and Daybreak are running out of time. Cheap models capable of dangerous hacking are now available to anyone, without the normal safeguards for refusing malicious actions. We need to fix vulnerabilities across the industry so that we aren't caught unawares. And for one of the first times in computing history, we have the ability to! We can use frontier LLMs that move faster than a human to find and fix these issues in the time we have left. The hard remaining part is deploying the fixes.

This probably sounds like nonsense words or hysterical overreacting to most people, so here's what that means:

"GLM" is a kind of LLM (AI). The GLM family is open-weight, which means anyone can download and run the models.

"flash" means that it is cheap and fast to run, compared to most "frontier" models. "cheap" is relative, but think around 5-15k USD in hardware to run it locally.

"frontier" here means that the LLM is "close to the frontier of what AI is currently able to achieve".

Project Glasswing and Daybreak are initiatives to use LLMs to fix security issues across the tech industry.

"malicious actions" includes things like hacking infrastructure and telling people how to build pipe bombs.

The rest of this post is about what makes me so sure this is an imminent threat, and what we can do in response.

GLM

GLM 5.3-flash can be downloaded and modified by anyone in the world.

... continue reading

Explore topics: glm 5.3-flash project glasswing daybreak open-weight models llm security