Wired just published an interesting story about political bias that can show up in LLM's due to their training. It is becoming clear that training an LLM to exhibit a certain bias is relatively easy. This is a reason for concern, because this can "reinforce entire ideologies, worldviews, truths and untruths” which is what OpenAI has been warning about.
Poker players and other human lie detectors look for “tells,” that is, a sign by which someone might unwittingly or involuntarily reveal what they know, or what they intend to do. A cardplayer yawns when he’s about to bluff, for example, or someone’s pupils dilate when they’ve successfully drawn to an insider straight.
As part of SecurityScorecard’s commitment to making the world a safer place, we are now the first and only security ratings platform to integrate with OpenAI’s GPT-4 system. With this natural language processing capability, cybersecurity leaders can find immediate answers and suggested mitigations for high-priority cyber risks.
Security teams are faced with relentless cyberattacks, and they cannot engineer defenses fast enough. SOC teams face limited visibility, insufficient context, and the inability to identify the threats that matter. Analysts are even more burned out, switching from tool to tool, frantically trying to make sense of what they are seeing.
We’ve had occasion to write about ChatGPT’s potential for malign use in social engineering, both in the generation of phishbait at scale and as a topical theme that can appear in lures. We continue to track concerns about the new technology as they surface in the literature.
Advancements in AI have led to the creation of generative AI systems like ChatGPT, which can generate human-like responses to text-based inputs. However, these inputs are at the discretion of the user and they aren’t automatically filtered for sensitive data. This means that these systems can also be used to generate content from sensitive data, such as medical records, financial information, or personal details.
The nature of an advanced artificial intelligence (AI) engine such as ChatGPT provides its users with an ability to use and misuse, potentially empowering both security teams and threat actors alike. I’ve previously covered examples of how ChatGPT and other AI engines like it can be used to craft believable business-related phishing emails, malicious code, and more for the threat actor.
We’re pleased to share that Salt has extended the capabilities of our powerful AI algorithms, further strengthening the threat detection and API discovery abilities of the Salt Security API Protection Platform. (Check out today’s announcement.) Here at Salt, we always look forward to the RSA Conference, but this year we are doubly excited to attend and showcase these new advanced capabilities! Salt invests significant resources into the continued innovation of our API security platform.
The Verge came out with an article that got my attention. As artificial intelligence continues to advance at an unprecedented pace, the potential for its misuse in the realm of information security grows in parallel. A recent experiment by data scientist Izzy Miller shows another angle. Miller managed to clone his best friends' group chat using AI, downloading 500,000 messages from a seven-year-long group chat, and training an AI language model to replicate his friends' conversations.
Not long after impressing Microsoft 365 customers with the recent Microsoft 365 Copilot announcement, Microsoft have launched another AI-powered Copilot product. This time with a whole new set of possibilities – introducing Microsoft Security Copilot.
ChatGPT may not be used by all organizations and may even be banned. But that doesn't mean you don't have exposure to the security risks it contains. This post looks at why ChatGPT should be part of your threat landscape.
Remember The Sims? Well Stanford created a small virtual world with 25 ChatGPT-powered "people". The simulation ran for 2 days and showed that AI-powered bots can interact in a very human-like way. They planned a party, coordinated the event, and attended the party within the sim. A summary of it can be found on the Cornell University website. That page also has a download link for a PDF of the entire paper (via Reddit).
The use of Large Language Models (LLMs) is the fine tuning AI engines like ChatGPT need to focus the scam email output to only effective content that results in a wave of new email scams.
AI coding is here, and it’s transforming the way we create software. The use of AI in coding is actively revolutionizing the industry and increasing developer productivity by 55%. However, just because we can use AI in coding doesn't mean we should adopt it blindly without considering the potential risks and unintended consequences.
With all the overwrought hype with ChatGPT and AI…much of it earned…you could be forgiven for thinking that only the bad actors are going to be using these advanced technologies and the rest of us are at their mercy. But this is not an asymmetric battle where the bad actors use AI and the rest of us are struggling using our pencils and abacuses to catch up. It is the good side that invented and is accelerating AI. It is the good scientists that made ChatGPT and all of its competitors.
Anticipation leads people to suspend their better judgment as a new campaign of credential theft exploits a person’s excitement about the newest AI systems not yet available to the general public. On Tuesday morning, April 11th, Veriti explained that several unknown actors are making false Facebook ads which advertise a free download of AIs like ChatGPT and Google Bard.
Getting the benefits of AI threat detection tools is becoming less of an option for security operation centres (SOCs). Last year, the UK experienced more cyber attacks than any other country in Europe. According to IBM’s X-Force Threat Intelligence Index report, nearly half (43%) of all cyber attacks in Europe targeted UK-based organisations.
When I wrote my first applications in high school, coding was a lot more time-consuming. I didn’t have libraries I could shop through with ready-made bits of code to drop in to save myself time or bridge the gap between the pieces of code I could write to create something more complex. Instead, I had to write everything from scratch. If I wanted to include any kind of image, I would have to draw it with mathematical coordinates, which took a ton of time.
In 1999, a far-fetched movie about a dystopia run by intelligent machines captured our imaginations (and to this day, remains my favorite film). Twenty-four years later, the line between fact and fiction has all but vanished and the blockbuster hits much differently. Are we entering the Matrix? Are we already in it? Can anyone be sure?
This MIT Technology Review headline caught my eye, and I think you understand why. They described a new type of exploit called prompt injection. Melissa Heikkilä wrote: "I just published a story that sets out some of the ways AI language models can be misused. I have some bad news: It’s stupidly easy, it requires no programming skills, and there are no known fixes.
Security AI usage has surged, and enterprises are reaping the benefits. In its 2022 Cost of a Data Breach Report, IBM found that organizations deploying security AI and automation incurred $3.05 million less on average in breach costs – the biggest cost saver found in the study. According to the study, organizations using security AI and automation detected and contained breaches faster. However, while leveraging AI clearly makes a difference, organizations must implement the right architecture.
When ChatGPT debuted in November 2022, it ushered in new points of view and sentiments around AI adoption. Workers from nearly every industry started to reimagine how they could accomplish daily tasks and execute their work — and the cybersecurity industry was no exception. Like shadow IT, new rogue AI tools — meaning AI tools that employees are adopting unbeknownst to the organization they work for — can pose security risks to your organization.
As a leader in applying AI to developer security, Snyk’s approach is unique. Today, we want to provide a glimpse at how Snyk currently uses AI and data science, as well as a sneak peek at what’s to come. Before diving in, we want to highlight two aspects of Snyk’s use of AI to set the stage.
In a groundbreaking move, Italy has imposed a ban on the widely popular AI tool ChatGPT. This decision comes in the wake of concerns over possible misinformation, biases and the ethical challenges AI-powered technology presents. The ban has sparked a global conversation, with many speculating whether other countries will follow suit.
New insights from cybersecurity artificial intelligence (AI) company Darktrace shows a 135% increase in novel social engineering attacks from Generative AI. This new form of social engineering that contributed to this increase is much more sophisticated in nature, using linguistics techniques with increased text volume, punctuation, and sentence length to trick their victim. We've recently covered ChatGPT scams and other various AI scams, but this attack proves to be very different.
Using the lure of ChatGPT’s AI as a means to find new ways to make money, scammers trick victims using a phishing-turned-vishing attack that eventually takes victim’s money. It’s probably safe to guess that anyone reading this article has either played with ChatGPT directly or has seen examples of its use on social media. The idea of being able to ask simple questions and get world-class expert answers in just about any area of knowledge is staggering.
