Understand the Risks, Enhance Your Security, and Safeguard Your Digital Life from Emerging AI Threats
ChatGPT Confirms Data Breach, Raising Security Concerns
So, ChatGPT got a taste of its own medicine. Instead of being used to launch cyberattacks, it’s now the target of a data breach. OpenAI, ChatGPT’s developer, admitted that a vulnerability in the code’s open-source library caused the issue. But hey, with great popularity comes great responsibility, right? This AI Swiss Army knife was a huge hit, getting 100 million monthly users in record time. But, with such fame, it’s only a matter of time before the baddies come knocking.
The breach allowed users to see other active users’ chat history. But dig a little deeper, and it turns out some folks could also see payment info, like names, email addresses, and the last four digits of credit card numbers. Oops.
This incident might be small potatoes, but it’s a wake-up call about the risks chatbots pose. They’re like airplanes’ black boxes, storing loads of data that become fair game for other users. And with AI getting better at creating realistic phishing emails and conspiracy theories, we’ve got a whole new world of cyber trouble ahead.
OpenAI’s working on preventing future breaches by offering bounties for finding vulnerabilities. But it seems they’re more focused on protecting the tech from outside attacks than stopping the chatbot from being the source of cyberattacks. Only time will tell if ChatGPT will play the victim or the villain in the cybersecurity world.
READ THE ARTICLE ON SECURITY INTELLIGENCE.
ChatGPT has generated a brand new app sector — and European devs are leading it
ChatGPT’s causing quite the hullabaloo, creating a new AI chat app sector, and Europe’s taking the lead. Since its November 2022 debut, these apps have been downloaded a whopping 23.6 million times. Europe’s churning out top apps like nobody’s business, with 14 of them snagging 8 million downloads, or 34% of the total.
The Americas and Asia are trailing behind, with 2.7 million and 7.4 million downloads, respectively. App Radar‘s bigwig, Thomas Kriebernegg, says this boom is one of the biggest he’s seen, and he credits Europe’s tech-savvy public for the rapid growth.
Turkey’s making a splash in the scene with 5.9 million downloads across four apps, but the cash hasn’t quite caught up with the hype. Most AI chat apps offer free basic use and rely on ads and subscriptions for income, but users are playing hard to get with their wallets.
Despite the money game, Kriebernegg thinks the sector will keep growing. Right now, most apps help with work stuff or offer a smarter search function, but as the tech evolves, we might see AI chat apps helping with creative tasks like making art or music. The future’s looking bright, folks!
READ THE ARTICLE ON THE NEXT WEB.
Microsoft could offer private ChatGPT to businesses for “10 times” the normal cost
Microsoft’s cookin’ up a privacy-focused ChatGPT for big shots like banks and healthcare providers who are itchin’ to keep their data locked up tight. This high-security version could be announced soon and would run on dedicated servers, keeping nosy AI out of sensitive info.
But there’s a catch – this private version could cost a pretty penny, like 10 times more than the regular ChatGPT. OpenAI’s also workin’ on something similar, but Microsoft’s got the edge by using its own Azure platform.
This whole shebang comes from a big ol’ investment Microsoft made in OpenAI, and it looks like they might even compete for the same customers. Some companies, like Samsung, Verizon, and a bunch of banks, have already given regular chatbots the boot, so they might be first in line for this fancy, private ChatGPT.
READ THE ARTICLE ON ARSTECHNICA.
MosaicML launches new service in bid to challenge OpenAI on price
MosaicML, a brainchild of ex-Intel bigwigs and smarty-pants academics, is throwin’ down the gauntlet against AI big dogs like OpenAI. Born in 2021, they’ve been helpin’ folks build custom AI systems on the cheap. Now, they’re stepping up their game by offering “inference services” to app developers, makin’ their apps even cooler with text and image tricks.
Their big claim? They’re 15 times cheaper than the competition, and still bringin’ home the bacon. MosaicML CEO Naveen Rao says they know their stuff just as well as the other guys. Their secret sauce? Foundation models (like those used by Microsoft and Google). But they’re shakin’ things up by lettin’ customers run their code on their own hardware, keepin’ their data safe and sound.
Rao believes customers dig their service ’cause they get to keep their precious data private, and own the models built on it. Clever move, MosaicML. Clever move.
CEO of Google’s DeepMind says we could be ‘just a few years’ from A.I. that has human-level intelligence
Google’s main AI research lab CEO Demis Hassabis believes artificial general intelligence (AGI) could rival human smarts in just a few years. AGI, a big ol’ deal for AI researchers, would understand the world as well as you and me. Hassabis says recent progress has been incredible, and he doesn’t see it slowing down. In fact, he thinks it might speed up.
AI has been the cool kid on the tech playground, and a lot of money is going towards generative AI, which creates text or images based on prompts. However, many researchers are already moving on to AGI, which would be more sophisticated and self-aware.
DeepMind, a Google subsidiary, has made impressive strides in AI, like predicting the structure of proteins and diagnosing complex eye diseases. With Hassabis as CEO, Google is merging its AI research team with DeepMind to focus on AGI.
But not everyone’s on board with fast-tracking AGI. Some AI experts worry it could lead to nuclear-level catastrophes or even a full-blown AI takeover that destroys humanity. Over 1,000 technologists, including Elon Musk and Steve Wozniak, signed a letter calling for a pause on advanced AI development to focus on ethics.
Hassabis promises that DeepMind’s AGI work won’t put civilization at risk, saying Google will develop AGI responsibly and cautiously, like a mad scientist who’s learned their lesson.
Geoffrey Hinton talks about the “existential threat” of AI
Geoffrey Hinton, big cheese of deep learning, is peacing out from his Google AI gig after 10 years. Why? He’s getting antsy about the potential dangers of AI and wants to speak his mind. Will Douglas Heaven from MIT Technology Review had a chat with Hinton about his concerns. Later, they talked at EmTech Digital, where Hinton said, “humanity could be just a passing phase in the evolution of intelligence.” Check out their full convo in the video below.