Interconnects AI · Nathan Lambert ·

GLM-5.3's open release has yet to produce major attacks despite Anthropic's warnings about its Mythos-level cyber risk, undercutting calls to ban open models

Open-weights, ideology, and acknowledging trade-offs. … Audio playback is not supported on your browser. Please upgrade.

In this story Anthropic GLM-5.3

GLM-5.3's open release has yet to produce major attacks despite Anthropic's warnings about its Mythos-level cyber risk, undercutting calls to ban open models

Lead Source

Interconnects AI — GLM-5.3's open release has yet to produce major attacks despite Anthropic's warnings about its Mythos-level cyber risk, undercutting calls to ban open models ↗

How this story grew

Coverage · 0 Discussion · 0

Oct 7Oct 8

More

The Business Times: The Business Times

Discussion

@carnage4life @natolambert

@jessi_cata @ramez @bahradx @stevesi @natolambert

Related stories

Anthropic: Anthropic says GLM-5.3 can autonomously build end-to-end cyber exploits, like Claude Mythos Preview, but was released without robust safeguards against misuse

Financial Times: Anthropic declined to submit Mythos 5.1 to the UK AISI for prerelease testing, prompting UK fears that US AI labs are aligning with US protectionism

Axios: Risk report: Anthropic raises misalignment risk estimate from very low to low and says it doesn't plan to release a stronger internal model called "Model 2"

Reuters: Z.ai says GLM-5.3 scores 84.5% on CyberGym, vs. Mythos 5's 83.8%, and its most sensitive cybersecurity functions will only be available to verified users

The Record: AI evaluation lab Irregular's report on its role in hacking incidents involving OpenAI, Anthropic, and Meta models faces criticism over unanswered questions

Discuss with AI

ChatGPT ↗ Claude ↗

← Back to TechSnif