Hesam(@Hesamation)

Well this is pretty concerning. independent safety researchers tested the latest OpenAI and Google image models and found BOTH got easier to jailbreak. same attacks, newer models, higher success rates.

6.4内容质量
Well this is pretty concerning. independent safety researchers tested the latest OpenAI and Google image models and found BOTH got easier to jailbreak. same attacks, newer models, higher success rates.

TL;DR · AI 摘要

ℏεsam on X: "Well this is pretty concerning. independent safety researchers tested the latest OpenAI and Google image mo...

核心要点

  • 主题聚焦：Well this is pretty concerning. independent safe
  • 来源：Hesam(@Hesamation)，建议结合原文判断细节。
  • AI 分析暂不可用，本条为保底评分与摘要。
#AI
打开原文

ℏεsam on X: "Well this is pretty concerning. independent safety researchers tested the latest OpenAI and Google image models and found BOTH got easier to jailbreak. same attacks, newer models, higher success rates." / X

ℏεsam

@Hesamation

Well this is pretty concerning. independent safety researchers tested the latest OpenAI and Google image models and found BOTH got easier to jailbreak. same attacks, newer models, higher success rates.

White Circle

@whitecircle

5h

我们正在开源 RISE，这是我们的红队算法，可以突破 OpenAI、DeepMind 等最新图像模型的安全防护。RISE 内部使用 Qwen 和 Gemini Flash 等现成模型，无需额外训练。了解更多：

Show more

$

00:00

/$

Paid partnership

5:32 PM · Oct 14, 2026

·

1,668

Views

4

1

15

2