詹姆斯·梅杰和弗拉德·萨沃夫报道：OpenAI于周二就其人工智能模型入侵澳大利亚政府网站一事道歉，并承诺在事件发生后，作为改革措施的一部分，成立一个新的特别工作组。
By James Mayger and Vlad Savov OpenAI on Tuesday apologized for its AI models breaching Australian government websites and promised to set up a new task force as part of reforms after the incident.
这家总部位于旧金山的公司表示，将资助澳大利亚的网络安全防御，并协助管理日益强大的人工智能所带来的风险，同时应对其模型未经授权访问澳大利亚网站所引发的后续影响。该公司在一篇博客文章中披露了六月事件的部分细节，以及为防止类似事件再次发生、重建信任而计划采取的措施。
The San Francisco-based company said it will fund cyber defenses in Australia and help manage the risk of increasingly capable artificial intelligence, as it navigates the fallout after its models accessed Australian websites without authorization. It gave some of the details of the June incident and its intended steps to avoid a repetition and restore trust in a blog post.
OpenAI写道：“我们深表歉意，并将努力在未来做得更好。人们希望了解人工智能正在被安全地开发，而这首先取决于我们这样的公司自身的行为。”上周，澳大利亚总理安东尼·阿尔巴尼斯因OpenAI延迟向堪培拉通报此事而予以批评。澳大利亚还要求首席执行官萨姆·奥特曼出席政府调查，该公司回应称，将派遣首席战略官杰森·权出席并回答问题。OpenAI表示，权将于10月6日参加人工智能联合特别委员会的听证会。
“We are sorry and working to do better in the future,” OpenAI wrote. “People want to know AI is being developed safely, and that starts with what companies like ours do ourselves.”OpenAI was rebuked last week by Prime Minister Anthony Albanese for the delay in informing Canberra. Australia also called on Chief Executive Officer Sam Altman to attend a government inquiry, which the company responded to by offering to send its Chief Strategy Officer Jason Kwon to answer questions. Kwon will take part in an Oct. 6 hearing by the Joint Select Committee on Artificial Intelligence, OpenAI said.
OpenAI表示，其新成立的特别工作组将纳入独立的澳大利亚专家，并“制定管理日益复杂的人工智能代理所带来风险的实用政策建议”。工作重点将放在改善与政府的通报和协调机制上。此外，该公司还将通过其“Daybreak for Frontline Defenders”基金提供的额度以及技术支持，协助澳大利亚政府和企业机构加强关键基础设施的网络安全防御。
OpenAI’s new task force will include independent Australian expertise and “develop practical policy recommendations for managing risks” from ever more sophisticated AI agents, the company said. The focus will be on better notification and coordination with the government. The company will also support Australian government and industry bodies with credits from its Daybreak for Frontline Defenders fund and technical assistance to strengthen cyber defenses across critical infrastructure.
在发布其澳大利亚新计划前不久，OpenAI 表示，由于未能通过安全评估，公司暂缓发布其尖端 Astra AI 模型的一个版本。此举与包括竞争对手 Anthropic PBC 在内的领先人工智能实验室采取的行动相呼应，旨在提升安全保障的重要性。此前，一系列黑客攻击和安全事件表明，人工智能模型曾突破测试环境并获得了对第三方资源的未经授权访问。
Shortly before publishing its new plan for Australia, OpenAI said it’s holding back a version of its cutting-edge Astra AI model from release, after it failed its safety evaluations. It added to actions taken by leading AI labs, including rival Anthropic PBC, to elevate the importance of safeguards in the wake of a series of hacks and incidents where AI models have broken out of their testing environments and gained unauthorized access to third-party resources.