Vitalik backs Kimi 3.1 for omitting top-end offensive cyber capabilities from training data
𝕏/@VitalikButerin •
Revision history
4 recorded changes
Want your article here?
Promote with Leviathan NewsVitalik Buterin said he supports Kimi 3.1's decision to exclude top-end offensive cyber capabilities from its training data. He argued that restricting use cases is far less harmful than restricting users, and that the world needs time to prepare—hardening systems, auditing old code, and formally verifying new code—before powerful AI-enabled cyber attacks arrive. He warned that mishandling this first wave of AI risk could lead to heavily locked-down models, KYC gating, pervasive surveillance, and tiered model access based on credentials and nationality.
AI summary of the linked source — not editorial content.
Story signal · how it spread on X · via Elfa
as ofSpread
i
Distinct X and Telegram accounts Elfa counted behind this story’s reports, summed over the thread. Not posts, not impressions.
6 accounts
first post
Leviathan +4.7 h later
LN
Thread · 1 event
- First reportVitalik backs limiting Kimi K3.1’s offensive cyber capabilities while warning against restrictive AI governance6 acc.
- Leviathan published this article
All times UTC.
Cited posts
- @BlockBeatsAsia BlockBeats｜We're hiring!First reportOn X ↗
Vitalik 谈 Kimi 3.1：AI 安全不该以牺牲开放性为代价。 他表示，可以接受 Kimi 3.1 在训练中有意排除最顶尖的网络攻击能力。随着 AI 发现漏洞、编写攻击代码的能力不断增强，社会需要时间修补旧系统、审查新代码，在风险全面爆发前做好准备。 但 Vitalik 更担心另一种结果：如果 AI… https://t.co/I3lMHKjTD9
- @PANewsCN PANews丨APP全面升级On X ↗
Vitalik声援Kimi K3.1删除顶级网络攻击能力：限制用途，好过限制用户 以太坊联合创始人Vitalik Buterin发文称，他赞同月之暗面在Kimi K3.1的训练数据中刻意剔除顶级网络攻击能力。… https://t.co/Tj5CsVZRRt
- @ChainCatcher_ ChainCatcherOn X ↗
Vitalik：支持 Kimi 3.1 限制高端网络攻击能力训练 Vitalik 发文表示，他支持 Kimi 3.1 在训练数据中主动排除顶级进攻性网络安全能力，认为“限制使用场景”远好于“限制用户”。 他指出，社会仍需要时间为更强大的 AI…
Show 3 more cited posts
- @sleepy0x13 sleepyOn X ↗
Vitalik 今天公开支持 Kimi 主动削弱模型能力。 月之暗面在最新的 Kimi K3.1 Pro 训练过程中，刻意剔除了部分网络安全相关数据和强化学习任务，试图限制模型的高级网络攻击能力。 Kimi 自己的测试显示，未经限制的版本在漏洞发现与利用任务上的成功率为 90.5%，最终发布版本降到了… https://t.co/zKCDXAXB3E
- @BlockTempo 動區動趨 BlockTempoOn X ↗
V 神挺 Kimi 3.1「不教進階駭客」：限制用途，好過限制人@VitalikButerin 表示，他可以接受 Kimi 3.1 刻意不把頂尖攻擊性網路能力放進訓練資料。… https://t.co/Rdp5ej307m
- @BSCNews BSCNOn X ↗
Vitalik Buterin Warns Against a Locked Down AI Future Ethereum (@ethereum) ethereum:native co-founder, Vitalik Buterin (@VitalikButerin), supports Kimi K3.1 Pro’s decision to limit offensive cyber capabilities. He argues restricting use cases is better than restricting…
Social-monitoring data, not investment advice Elfa post text from Xas of
Explore the topic
Loading related coverage…
Comments
Loading comments…
