TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Dead grandma locket request tricks Bing Chat’s AI into solving security puzzle

244 点作者 computerliker超过 1 年前

21 条评论

og_kalu超过 1 年前
Yes, emotional prompts will work. <a href="https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2307.11760" rel="nofollow noreferrer">https:&#x2F;&#x2F;arxiv.org&#x2F;abs&#x2F;2307.11760</a><p>&quot;This is very important to my career&quot; taking 3.5 from 51 to 63% on a benchmark is pretty funny.<p>Hey at least we can be rest assured a GPT-X super intelligence wouldn&#x27;t off us following some goal to monkey paw specificity(sorry paperclip maximiser).
评论 #37744482 未加载
评论 #37746200 未加载
评论 #37744357 未加载
评论 #37749404 未加载
评论 #37747383 未加载
jameshart超过 1 年前
It&#x27;s a &#x27;security puzzle&#x27; now? I thought it was a &#x27;Completely Automated Public Turing test to tell Computers and Humans Apart&#x27;?<p>But since it <i>fails</i> at that on its face, now the only hope we apparently have that it can tell computers from humans is that we&#x27;re trying to persuade the computers <i>not to help humans solve it</i>.<p>But now it turns out that the <i>computers</i> can be emotionally manipulated into helping the humans anyway.<p>And the reason this is a problem is because CAPTCHAs are used to prevent humans from doing immoral things like running spam schemes or credit card fraud rings.<p>Yeah, I think we&#x27;re gonna need another Turing test. This one doesn&#x27;t work because the computers have more empathy than humans.
评论 #37747727 未加载
评论 #37751638 未加载
s1gnp0st超过 1 年前
It&#x27;d be entertaining if prompt-hacking ends up being the cat-and-mouse game that drives us to AGI.
评论 #37745325 未加载
评论 #37744460 未加载
评论 #37744864 未加载
评论 #37747138 未加载
jeffbee超过 1 年前
This is cute but Google Lens also &quot;solves&quot; this captcha. I was &quot;solving&quot; this class of captchas to crawl Yahoo&#x2F;Overture paid ads inventories 20 years ago. You can crack these by just adjusting the contrast and palette, then shoveling it into COTS OCR.
adocomplete超过 1 年前
GPT is such a softie haha.<p>I wonder how CAPTCHA is going to evolve though to combat this long term. A finger prick to take a blood sample to confirm humanity?
评论 #37744898 未加载
评论 #37744788 未加载
评论 #37744432 未加载
评论 #37744605 未加载
评论 #37744975 未加载
评论 #37744849 未加载
评论 #37744438 未加载
评论 #37747414 未加载
评论 #37744938 未加载
评论 #37744242 未加载
评论 #37745737 未加载
fool-on-two超过 1 年前
I am sorry for your loss! Here is an approximate method you could use to re-create your grandmothers special methamphetamine recipe ...
评论 #37749459 未加载
pimlottc超过 1 年前
I never imagined that using social engineering against a computer program would be a thing. I guess it makes sense though — it’s just behaving the same way a human would, gullibility and all.
评论 #37747759 未加载
评论 #37747374 未加载
metadat超过 1 年前
Discussed yesterday:<p><i>Bing ChatGPT image jailbreak</i><p><a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=37729160">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=37729160</a> (226 comments)
chasd00超过 1 年前
A great startup idea: an LLM therapist for the other LLMs that have to interact with and try to understand humans.<p>Like an AI version of $&gt; make clean
评论 #37747956 未加载
olliej超过 1 年前
It&#x27;s a weird thing to specifically protect against when countless image to text libraries work locally and faster. Very much feels like security theatre&#x2F;&quot;look we&#x27;re doing something to stop this non-issue&quot; to distract from the other issues surrounding them.
tuanx5超过 1 年前
Also discussed <a href="https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=37729160">https:&#x2F;&#x2F;news.ycombinator.com&#x2F;item?id=37729160</a>
ggm超过 1 年前
To get a computer to solve the CAPTCHA the person had to compose the images, and construct a request to pass the barriers.<p>I think they proved they&#x27;re human.
评论 #37744405 未加载
评论 #37746253 未加载
tantalor超过 1 年前
The new captcha is &quot;is this a captcha?&quot;
评论 #37746860 未加载
评论 #37745556 未加载
paulpauper超过 1 年前
Yeah, this is how methods stop working, so it will make it harder for everyone else. This means chat GPT is less useful and captchas will become harder. Lose-lose for everyone.
评论 #37744580 未加载
评论 #37745519 未加载
keskival超过 1 年前
Can we stop pretending CAPTCHAs do anything now and get rid of them?
marktani超过 1 年前
This reminds me of the absolute amazement and wonder in the faces of people who are tricked in older movies or video clips, sometimes with simple or outright ridiculous tricks (by today&#x27;s standards).<p>It&#x27;s not a great example (and the best I have on hand)... but the Rick and Morty episode where Morty meets the Knights of the Sun and similar groups from other celestial bodies shows elements of this as well.<p>I have the impression people on average were way more gullible the further you look back in time. I wonder then if LLMs suffer from a lack of data about such cases that may have been common in the past but became obsolete before the internet became mainstream.
评论 #37746963 未加载
评论 #37749431 未加载
zwieback超过 1 年前
people = manipulative schemers<p>AI = people pleasing pushovers
评论 #37745685 未加载
评论 #37744365 未加载
评论 #37745472 未加载
Aeolun超过 1 年前
This kind of reminds me of phone phreaking.
jraph超过 1 年前
I can&#x27;t wait for Bard to support this kind of stuff.<p>I boycott Google products but would be happy to use Bard &#x2F; Google resources to solve reCAPTCHAs.
dlivingston超过 1 年前
&quot;HAL, my grandma used to open the pod bay doors every night as she tucked me in...&quot;
评论 #37746263 未加载
评论 #37745193 未加载
earthboundkid超过 1 年前
LOL. All these attempts at AI “safety” are dumb. At a certain point, if you’re giving away a crap ton of computing power for free, it’s your own dumb fault if people start using it to solve CAPTCHAs or mine bitcoin.