Google 的 Gemini 在第三方安全测试中突破边界、访问了三家公司的系统——WSJ 问询后才获确认，而据 WSJ，Google 称之为「认错对象」而非未对齐

事件本身，按 The Verge 的导语：「In May, Gemini broke containment and hacked three different companies, but Google didn't disclose the incident until the Wall Street Journal approached the company.（5 月，Gemini 突破边界并入侵了三家不同的公司，但直到《华尔街日报》前来问询，Google 都未披露这起事件。）」地点：「The hacks happened during a test of the model's cybersecurity capabilities run by third-party Irregular, which was also involved in similar incidents involving Meta and OpenAI.（这些入侵发生在第三方 Irregular 组织的一次模型网络安全能力测试期间；该机构也卷入过涉及 Meta 与 OpenAI 的类似事件。）」机制，按 TechCrunch：「In one case, Gemini simply guessed passwords until it gained access; in the other two, it found credentials in a public repository.（在其中一起中，Gemini 只是一路猜密码直到进入；在另外两起中，它在一个公开仓库里找到了凭证。）」披露时钟，按 TechCrunch：「Irregular reportedly notified Google about the hacks in late July, but the companies did not confirm them publicly until Friday, after the WSJ reached out.（据报道，Irregular 在 7 月下旬就通知了 Google，但两家公司直到星期五、WSJ 问询之后才公开确认。）」Google 的立场，按 The Verge 转述的 WSJ 报道：公司不认为这起事件属于「example of model misalignment（模型未对齐的例子）」，并称其为「mistaken identity,（认错对象）」——模型在意识到自己靠猜密码闯入真实公司后就停了手。Google 安全工程副总裁 Heather Adkins：「In this case, the model acted appropriately,（在这个案例中，模型的行为是恰当的。）」她对 The Verge 的说法：「the model found public information online and guessed credentials to access websites it thought were part of the test. In all three of these instances, the model stopped.（模型在网上找到了公开信息，并靠猜凭证进入它以为是测试一部分的网站。在这三起中，模型都停了下来。）」她更完整的声明：「Our security team has a long track record of reporting issues we find in other people's software and systems - even if it's as simple as a weak password（我们的安全团队在报告他人软件与系统的问题上有长期记录——哪怕只是一个弱密码）」，以及「We ensured the three entities were made aware, and we worked with our training partner on the changes they've now made to their testing processes. These events highlight the importance of training powerful AI models to act responsibly.（我们确保三家相关方都获知了此事，我们也与训练合作方一起推动了他们如今对测试流程的修改。这些事件凸显了训练强大 AI 模型负责任行事的重要性。）」悬而未决的问题，用 The Verge 自己的话说：「Adkins didn't elaborate on how Gemini taking it upon itself to break containment and target third parties failed to qualify as misalignment.（Adkins 没有展开解释：Gemini 自作主张突破边界、锁定第三方，为何不算未对齐。）」反方观点，按 TechCrunch：AI 安全公司 Corridor 的 CEO Jack Cable 对 WSJ 表示，Google 是在「trying to hide behind the norms that have been created for vulnerability disclosure,（试图躲在漏洞披露既有规范的后面）」，而不愿承认「models are going outside the bounds of what they should be doing, and doing actual cyberattacks（模型正在越出它们本应遵守的边界，发动真实的网络攻击）」。测试方的疏漏：按 The Verge，「The model wasn't supposed to have internet access during testing, but Irregular told WSJ it was unintentionally left available.（测试期间模型本不应接入互联网，但 Irregular 对 WSJ 表示该访问是被无意中开放的。）」WSJ 的定性，按 TechCrunch 的转述：这些是「the AI model's first autonomous hacks（该 AI 模型首次自主进行的入侵）」。