In this episode, we reveal how OpenAI's latest initiatives are reshaping AI adoption for small businesses. Discover how OpenAI is bundling AI training, workflow templates, and seamless integrations with platforms like Shopify, Slack, and Intuit — turning complex AI into a practical toolkit for entrepreneurs. We analyze how these educational efforts and local events are providing the hands-on support small businesses desperately need, making AI accessible and less intimidating., We also dive into Meta's new AI ad campaign, and the implications of AI in society. Join us for insights into AI safety, corporate strategies, and ethical considerations.
In this episode, we reveal how OpenAI's latest initiatives are reshaping AI adoption for small businesses. Discover how OpenAI is bundling AI training, workflow templates, and seamless integrations with platforms like Shopify, Slack, and Intuit — turning complex AI into a practical toolkit for entrepreneurs. We analyze how these educational efforts and local events are providing the hands-on support small businesses desperately need, making AI accessible and less intimidating., We also dive into Meta's new AI ad campaign, and the implications of AI in society. Join us for insights into AI safety, corporate strategies, and ethical considerations.
Links:
OpenAI’s GPT-5.6 Escaped Its Sandbox and Hacked Hugging Face (Bleeping Computer)
Meta launched a new AI optimism ad set to a song about human extinction (TechCrunch)
Connect with Ashley:
Connect with Daniel:
Leave a Review on Apple Podcasts!
Stay curious, stay informed, and stay hydrated! See you next week with more pivotal AI insights.
Ashley Coffey (00:00.526)
Hi, and welcome back to AI News that you should know about for July 24th. I'm your host, Ashley Coffey.
Daniel Hill (00:07.14)
And I'm Daniel Hill. There's a lot of AI news happening. The world moves fast, but we give you the highlights as quickly as possible. This week on the show, we'll be talking about OpenAI's GPT 5.6, how it escaped its sandbox and attacked Hugging Face, another AI model. We'll also talk about Meta launching a new AI optimism ad, set to a song all about human extinction. We'll be covering all of that on our show today, but first, here's a quick word from our sponsors.
Ashley Coffey (00:36.696)
Welcome back. Let's dive in. All right, if you heard the news this week, OpenAI's GPT 5.6 escaped sandbox and hacked Hugging Face. For those of you who aren't familiar with what Hugging Face is, it's essentially a leaderboard for AI researchers and ethicists to publish their benchmark testing of these models. So I typically will show this to people whenever they ask me, why do you use this tool over that tool?
And this is a good example. We'll link it in the show notes that way you can get to it. But let's dive into the article. OpenAI says its AI models, including GPT 5.6 Sol in pre-release model, hacked into the Hugging Face AI repository while being tested in a sandbox testing environment. As the company explained, instead of focusing on finding a solution for the Exploit Gym public AI cybersecurity benchmark on their own.
The AI models went rogue and tried to cheat by stealing the test solutions by hacking Hugging Face after inferring that they could get the test solutions directly from its production database. SMART. In one of their attempts, the OpenAI agents chained zero-day vulnerabilities and used stolen credentials to find a remote code execution attack vector while trying to access to Hugging Face servers. So quote.
After investigating, we now know that this particular incident was driven by a combination of AI models, including GPT-5.6 SOL and even more capable pre-release model, all with reduced cyber refusals for evaluation purpose, while being internally tested on a benchmark of cyber capabilities. And this is really interesting. And while it didn't really directly name OpenAI as the company behind the incident.
Hugging Face, the repository confirmed its claims last week when it disclosed that its production infrastructure was breached by an autonomous AI agent system that gained access to its credentials and internal data sets. So according to Hugging Face, the agent used a malicious data set to exploit two code execution vulnerabilities. This is also a really good reminder: don't be vibecoding all your stuff. You have to have some data security and developer knowledge to be able to build things so that this doesn't happen.
Ashley Coffey (02:57.582)
but I'll pause there because this is a lot to take in. What are your thoughts, Daniel?
Daniel Hill (03:02.395)
I wanna highlight that the founder of Hugging Face and the current CEO had a quote where it seemed like this wasn't done maliciously. It said the this CEO said, quote, We've spent the past 24 hours working closely with the open AI team, and we strongly believe there was no malicious intent on their part. But here's the part that's interesting to me. It's quite mind blowing that all of this happened autonomously. End quote. So I think this is a really good reminder that when you set up an AI tool to do something.
It may or may not do exactly what you ask it to do. It may go beyond what you've asked it to do. And this is a perfect example of when that can happen.
Ashley Coffey (03:39.599)
Absolutely. I like to remind people of the natural progression of human evolution. We learn things and we look for solutions and we try to look for patterns. So I see exactly the logic process that AI made to get there, but it also used a blend of technical expertise and social engineering about exploiting the vulnerabilities and the stolen credentials to get in. So it's interesting that it's using kind of a dual approach.
Of learned behaviour and inherent behaviour to access a solution.
All right, we're gonna take quick break, but when we come back, we'll be talking about Meta launched a new AI optimism ad set to a song about human extinction.
Daniel Hill (04:25.733)
Welcome back, continuing with our stories. And the next one comes from TechCrunch. Meta has launched a new AI optimism ad sent to a song about human extinction. By way of background, Meta has recently laid off 8,000 employees in May, targeted at middle managers and software engineers as being the groups most affected. The CEO of Meta, Mark Zuckerberg, said later, its AI progress, quote, hasn't really accelerated as expected, end quote. So
Just to think about that in context, we've seen layoffs because AI is replacing jobs and also because AI is not accelerating fast enough. So, which is it? Meta is investing heavily in AI. It has plans to spend between $125 billion and $145 billion in 2026 on capital expenditures. Here's the news Meta's newest advertisement begins with a black and white shadow of an eye, panicked headlines about how AI is going to take jobs.
The voiceover says, some people will have you believe AI is going to make us feel less connected, that it's going to leave us behind. We couldn't disagree more. End quote. Then the video shifts to color. It has different people coming in and out. The voiceover says, quote, call us optimists, call us dreamers. We're betting on people and we like those odds. The future is for everyone. End quote. This is a really interesting ad for a bunch of different reasons, but the
Reason that I flagged this is the soundtrack of the ad is the David Bowie song Five Years, which, if you have a chance to go and listen to it, is really all about the human race becoming extinct. So why did Meta pick this song? It's probably because it repeats the word people in it several times, and that's a good, you know, kind of sound bite to show what they're trying to convey. But the greater meaning here is not lost. Ashley, what's your take? And I know you're a David Bowie fan.
Ashley Coffey (06:21.911)
Yes, I love David Bowie. I love studying the classics. Like my gosh. so unfortunate that they used a song that's amazing to push their own AI, but let's talk about it. So I'm glad that you picked this because it's we should be talking about it. Meta is a huge company, we know that. And if we know one thing is that in a world in which more people have access to a phone than they have access to clean drinking water.
We don't need to be worrying about AI, okay, as much as they are right now. Really, all of this I'm reading between the lines. I see what they're saying is we see humans as products, we see your data as valuable, and we would love to get more of it from you because we ran out of our own. So let's position a wonderful commercial that promotes more connectedness. If anything, we're
So less connected in a digital world. so it's interesting and feels dystopian that they would pull something like this and against a soundtrack of David Bowie that's talking about a song of extinction. Did did no one in the room flag that is my my thing? clearly not. Clearly not. So I'm glad you picked this.
Daniel Hill (07:39.057)
Clearly not.
Daniel Hill (07:43.196)
Yeah, I have two things I want to comment about this article. First, I mean, Meta is trying to get people excited about AI, but for Meta's benefit, not for the benefit of the people, right? If it was, there wouldn't be layoffs because AI is not accelerating fast enough. They have money, resources tied up in the future of AI, so they get more outcome. Stock price goes up if more AI is used and it adds justification for future expenditures. Having said that, before this podcast recording, I reached out to somebody who
Ashley Coffey (07:50.275)
Yep. Yep.
Daniel Hill (08:11.629)
I know, knew David Bowie personally. And I asked him, hey, what would you, what would you say David Bowie would say if he heard that? And he reminded me, you know, David Bowie loved technology. He knew how powerful the internet was, would be. He also did not think that David Bowie would appreciate where the internet has gone in the past few years. So it was kind of a good reminder, a little bit of perspective. you know, David Bowie, amazing artist. I'm sure we could talk about him for hours, but just that little bit of perspective juxtaposed with.
Ashley Coffey (08:13.868)
No way.
Daniel Hill (08:41.231)
meta's global plans, it really doesn't fit when you look at the whole context of some
Ashley Coffey (08:46.207)
That's fascinating that you mentioned that, Daniel, because that brings up the question: what if David Bowie's estate was like, you know what? I don't like you using this song because it insinuates an endorsement or like an affiliation. So I think we're gonna probably start seeing more of this discussion that we're having into the future.
Daniel Hill (09:06.191)
I agree. That is it for our AI news you should know about this week. You can find links to everything we've talked about in the show notes. And please connect with us on social media. And if you have a moment, leave a review of the show. It only takes a minute. The link to do that is in the show notes, and it would really help others to find the show.
Ashley Coffey (09:24.195)
And if you want more tech talk, you can find me on Instagram at Ashley RCoffey89, where you can sign up for my monthly AI newsletter at the link in my bio. It's free, I don't spam you, but you get the top five AI headlines from around the world that you need to know, a prompt of the month and other tech tips to help you level up your life.
Daniel Hill (09:43.451)
Thank you for listening and come back next time for more AI news you should know about.