AI maths advances raise fears over security of global encryption - FT中文网
登录×
电子邮件/用户名
密码
记住我
请输入邮箱和密码进行绑定操作:
请输入手机号码,通过短信验证(目前仅支持中国大陆地区的手机号)。如果该手机号尚未注册,将自动创建 FT中文网账号。短信可能需要几分钟送达,验证码15分钟内有效,请耐心等待:
请您阅读我们的用户注册协议和隐私权保护政策,点击下方按钮即视为您接受。
商业快报

AI maths advances raise fears over security of global encryption

AI’s growing mathematical prowess forces rethink of cyber security preparations long dominated by quantum threat
00:00

{"text":[[{"start":10.52,"text":"Advances in AI threaten to crack cryptographic tools used to shield digital data, triggering a scramble to ensure mainstream existing methods and planned next-generation safeguards are secure."}],[{"start":22.5,"text":"Regulators and institutions such as the UK’s Financial Conduct Authority are increasingly warning companies to make plans to switch the encryption methods used to protect sensitive information online."}],[{"start":33.12,"text":"Those calls have become heightened after OpenAI triggered a row by unveiling hundreds of mathematical papers, including “proofs” for some longstanding problems which it said included solutions generated by its AI models."}],[{"start":45.86,"text":"The Association for Human Mathematics, a group set up to protect the field against the “threat of artificial intelligence”, attacked the ChatGPT maker over the work and urged mathematicians to stop working with the company. OpenAI has defended its actions and pledged $5mn to a maths-focused foundation."}],[{"start":64.56,"text":"Experts said the AI-assisted breakthroughs raise questions about whether the technology could soon crack core existing cryptography methods, such as those used to stop hackers breaking into online bank accounts and databases."}],[{"start":77.28,"text":"“What’s going on in these [AI] systems is a complete kind of restructuring of what difficulty is — and difficulty underlies a lot of modern cryptography,” Australian mathematician Geordie Williamson said in a recent lecture hosted by Imperial College London. “I think that we should expect shocks in the future . . . very unsettling major, major shocks.”"}],[{"start":97.32,"text":"While AI has yet to break a core cryptography method and some cyber security experts remain sceptical it will do so, mathematicians warn the dangers are rising."}],[{"start":106.96,"text":"One reason is that the AI performance improvements in algorithmic design and number theory — relating to the properties of whole numbers — have increased the capacity to solve problems previously considered very hard."}],[{"start":118.68,"text":"One widely used cryptographic method is based on multiplying together two very large prime numbers. This has been an effective process until now, as it is easy for existing computers to do the multiplication, but harder to do the reverse process of working out what the pair of source numbers are."}],[{"start":134.66,"text":"AI advances are already having an impact on cyber security. Last week, companies across the world sought to quickly fix flaws in the widely used website programming language PHP, revealed in part by AI models developed by OpenAI and Google."}],[{"start":149.32,"text":"AI developments in maths have also thrown into doubt the previous assumption that the most pressing threat to cryptography is the advent of quantum computers — machines designed to exploit the physics of subatomic particles to achieve results beyond the capability of “classical” computers."}],[{"start":166.6,"text":"However, most experts believe that a sufficiently powerful quantum computer remains years from completion."}],[{"start":173.28,"text":"“What’s the crocodile that’s closest to the boat when it comes to our cryptographic algorithms?” said one finance industry insider. “Is it quantum or is it AI?”"}],[{"start":183.32,"text":"Cyber threats are “evolving quickly as AI and quantum computing advance”, the UK’s FCA told the FT. The regulator said it was trialling a new supervisory tool with 50 large institutions to assess their preparedness for an era of new technological risks to cryptography."}],[{"start":199.64,"text":"“Firms shouldn’t wait to act,” said the FCA, which plans to share practical guidance next month for how the more than 35,000 financial services companies it supervises in the UK should tackle the risks of AI and quantum computing."}],[{"start":213.64,"text":"“They should understand how their systems are protected today and start preparing now for a quantum-secure future,” it said."}],[{"start":220.84,"text":"The prospect of AI cracking encryption tools has emerged as a “significant risk” for “more mathematical forms of cryptography”, said Daniel Shiu, cryptomathematics professor at the University of Manchester."}],[{"start":232.8,"text":"The vulnerabilities included some of the “less mature” algorithms among a suite of new ones designed to be safe from assault by quantum computers, he said."}],[{"start":240.86,"text":"AI’s increasing mathematical aptitude was triggering action by authorities, Shiu said. It had led to a re-evaluation of some existing and proposed cryptographic tools in countries including the US, China and Germany, he said."}],[{"start":254.64,"text":"Telecoms group BT is among the companies that said it was already preparing to migrate to new encryption methods due to the future threat from quantum computers. The company added it was focused on ensuring cryptographic technologies can be “changed more readily as standards, technologies and threats evolve”."}],[{"start":276.36,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1793171229_3147.mp3"}

版权声明:本文版权归FT中文网所有,未经允许任何单位或个人不得转载,复制或以任何其他方式使用本文全部或部分,侵权必究。

为什么美国“有产者”也感受到了经济压力?

马斯特斯:从房屋保险到草坪养护,各项住房成本都在上涨,导致近半数美国房主在财务上捉襟见肘。

“希望这一切赶紧结束”:选举广告狂轰滥炸让美国选民厌倦

距离中期选举仅剩一周,两党候选人密集投放攻击性广告。而承受广告狂轰滥炸的是日益恼火的选民。

投资机构能大到不必在乎道德准则吗?

挪威政府正准备就其规模2.3万亿美元的主权财富基金在投资过程中应如何处理道德准则问题提交提案。

Lex专栏:Anthropic、Meta与K型公司

有影响力的股东与毫无影响力的股东之间的差距正在扩大。

iPhone Duo表现如何?

HTSI科技评测员称,这是“我见过的迄今为止最漂亮的折叠设备”。

FT社评:对冲AI热潮的局限

随着科技投资激增,投资者寻求保护的行动变得愈发迫切,也愈发困难。
设置字号×
最小
较小
默认
较大
最大
分享×