Researchers develop technique to prevent software bugs

A team of computer scientists led by the University of Massachusetts Amherst recently announced a new method for automatically generating whole proofs that can be used to prevent software bugs and verify that the underlying code is correct.
The effects of buggy software can range anywhere from the annoying-glitchy formatting or sudden crashes-to potentially catastrophic security breaches or the precision software used for space exploration or controlling healthcare devices.
Of course, there have been methods for checking software for as long as it has existed.
One popular method is the simplest: you have a human being go through the code, line by line, manually verifying that there are no errors.
Or you can run the code and check it against what you expect it to do.
The problem with both methods is that they are prone to human error, and checking against every possible glitch is extraordinarily time-consuming, expensive, and infeasible for anything but trivial systems.
A much more thorough but harder method is to generate a mathematical proof showing that the code does what it is expected to do and then use a theorem prover to ensure the proof is also correct.
Manually writing these proofs is incredibly time-consuming and requires extensive expertise.
With the rise of LLMs, of which ChatGPT is the most famous example, a possible solution is to try to generate such proofs automatically.
The work was done as a collaboration with Google, and built on top of a significant amount of prior research.
First, whose team performed its work at Google, used Minerva, an LLM trained on a large corpus of natural-language text, and then fine-tuned it on 118GB of mathematical scientific papers and webpages containing mathematical expressions.
Next, she further fine-tuned the LLM on a language, called Isabelle/HOL, in which the mathematical proofs are written.
Baldur then generated an entire proof and worked in tandem with the theorem prover to check its work.
When the theorem prover caught an error, it fed the proof, as well as information about the error, back into the LLM, so that it can learn from its mistake and generate a new and hopefully error-free proof.
This process yields a remarkable increase in accuracy.
The tool for automatically generating proofs is called Thor, which can generate proofs 57% of the time.
When Baldur is paired with Thor, the two can generate proofs 65.7% of the time.
Though there is still a large degree of error, Baldur is by far the most effective and efficient way yet devised to verify software correctness, and as the capabilities of AI are increasingly extended and refined, so should Baldur's effectiveness grow.
In addition to First and Brun, the team includes Markus Rabe, who was employed by Google at the time, and Talia Ringer, an assistant professor at the University of Illinois-Urbana Champaign.
This work was performed at Google and supported by the Defense Advanced Research Projects Agency and the National Science Foundation.


This Cyber News was published on www.helpnetsecurity.com. Publication date: Wed, 10 Jan 2024 04:13:05 +0000


Cyber News related to Researchers develop technique to prevent software bugs

Researchers Uncover Simple Technique to Extract ChatGPT Training Data - Can getting ChatGPT to repeat the same word over and over again cause it to regurgitate large amounts of its training data, including personally identifiable information and other data scraped from the Web? The answer is an emphatic yes, according to ...
11 months ago Darkreading.com
What Is Software Piracy? - Software piracy has become a worldwide issue, with China, the United States and India being the top three offenders. In 2022, 6.2% of people worldwide visited software piracy websites. Software piracy doesn't require a hacker or skilled coder. Any ...
11 months ago Pandasecurity.com
Researchers extract RSA keys from SSH server signing errors - A team of academic researchers from universities in California and Massachusetts demonstrated that it's possible under certain conditions for passive network attackers to retrieve secret RSA keys from naturally occurring errors leading to failed SSH ...
11 months ago Bleepingcomputer.com
Purple teaming and the role of threat categorization - Red team assessment, penetration testing, and even purple team assessments are all designed to answer these questions. As attacks get more complex, these assessments struggle to provide comprehensive answers. These assessment services typically test ...
10 months ago Helpnetsecurity.com
'GhostRace' Speculative Execution Attack Impacts All CPU, OS Vendors - Researchers at IBM and VU Amsterdam have developed a new attack that exploits speculative execution mechanisms in modern computer processors to bypass checks in operating systems against what are known as race conditions. The attack leverages a ...
8 months ago Darkreading.com
DNS Tunneling Abuse Expands to Tracking & Scanning Victims - Attackers are taking malicious manipulation of DNS traffic to the next level, abusing DNS tunneling to scan a victim's network infrastructure as well as track victims' online behavior. Researchers from Palo Alto Networks' Unit 42 have identified ...
6 months ago Darkreading.com
Patch Now: Critical Windows Kerberos Bug Bypasses Microsoft Security - Microsoft eased enterprise security teams into 2024 with a relatively light January security update consisting of patches for 48 unique CVEs, just two of which the company identified as being of critical severity. For the second straight month, ...
10 months ago Darkreading.com
Google Researchers' Attack Prompts ChatGPT to Reveal Its Training Data - A team of researchers primarily from Google's DeepMind systematically convinced ChatGPT to reveal snippets of the data it was trained on using a new type of attack prompt which asked a production model of the chatbot to repeat specific words forever. ...
11 months ago 404media.co
GrimResource Technique Exploits Years-Old Unpatched Windows XSS Flaw - New GrimResource technique exploits a 2018-old, unpatched, Windows XSS flaw and crafted MSC files to deploy malware via the Microsoft Management Console. Researchers detected the new exploitation technique in the wild on June 6th, 2024. Exploiting ...
4 months ago Heimdalsecurity.com
The Crucial Need for a Secure Software Development Lifecycle in Today's Digital Landscape - In today's increasingly digital world, software is the backbone of business operations, from customer-facing applications to internal processes. The rapid growth of software development has also made organizations more vulnerable to security threats. ...
10 months ago Cyberdefensemagazine.com
Researchers develop technique to prevent software bugs - A team of computer scientists led by the University of Massachusetts Amherst recently announced a new method for automatically generating whole proofs that can be used to prevent software bugs and verify that the underlying code is correct. The ...
10 months ago Helpnetsecurity.com
Breaking Down the Blank Image Attack, a Strategy that Allows Malware to Evade Anti-Malware Software - Recent advances in internet security have allowed anti-malware software to block malware attacks more effectively. However, some malware developers employ techniques such as the “blank image attack” to bypass traditional anti-malware detection ...
1 year ago Hackread.com
Days After Google, Apple Reveals Exploited Zero-Day in Browser Engine - Apple has patched an actively exploited zero-day bug in its WebKit browser engine for Safari. Actively Exploited Apple yesterday described the vulnerability as something an attacker could exploit to execute arbitrary code on affected systems. ...
9 months ago Darkreading.com
Consumer Software Security Assessment: Should We Follow NHTSA's Lead? - The US National Highway Traffic Safety Administration is dedicated to its mission: "To save lives, prevent injuries, and reduce economic costs due to road traffic crashes, through education, research, safety standards, and enforcement." Is it time to ...
11 months ago Darkreading.com
Researchers automated jailbreaking of LLMs with other LLMs - AI security researchers from Robust Intelligence and Yale University have designed a machine learning technique that can speedily jailbreak large language models in an automated fashion. Their findings suggest that this vulnerability is universal ...
11 months ago Helpnetsecurity.com
Microsoft Gives Admins a Reprieve With Lighter-Than-Usual Patch Update - In what's sure to be a refreshing break for IT and security teams, Microsoft's monthly security update for December 2023 contained fewer vulnerabilities for them to address than in recent months. The update included fixes for a total of 36 ...
11 months ago Darkreading.com
Exposed Hugging Face API tokens jeopardized GenAI models - Lasso Security researchers discovered 1,681 Hugging Face API tokens exposed in code repositories, which left vendors such as Google, Meta, Microsoft and VMware open to potential supply chain attacks. In a blog post published Monday, Lasso Security ...
11 months ago Techtarget.com
Hackers Increasingly Use Microsoft OneNote to Deliver Malware - OneNote documents are increasingly being used by threat actors to send malware to unsuspecting end users via email, according to Proofpoint researchers. It infects victims with remote access malware that can be used to install additional malware, ...
1 year ago Cybersecuritynews.com
Revolutionizing WordPress Bug Bounty and Security: Latest Enhancements to the Wordfence Bug Bounty Program - Our team has triaged around 2,140 vulnerability submissions, with about 1,320 deemed in-scope. Together with our researchers and software vendors, we've protected millions of websites from vulnerabilities - and this is just the beginning. We're ...
6 months ago Wordfence.com
Understanding SBOMs - In recent years, the adoption of open-source software in development has surged, now comprising up to 90% of what's built. There is a crucial aspect to consider when integrating open-source software components. To make sure their software is safe, ...
11 months ago Securityboulevard.com
macOS Malware Mix & Match: North Korean APTs Stir Up Fresh Attacks - North Korean advanced persistent threat groups are mixing and matching components of two recently unleashed types of Mac-targeted malware to evade detection and fly under the radar as they continue their efforts to conduct operations at the behest of ...
11 months ago Darkreading.com
Meta AI Models Cracked Open With Exposed API Tokens - Researchers recently were able to get full read and write access to Meta's Bloom, Meta-Llama, and Pythia large language model repositories in a troubling demonstration of the supply chain risks to organizations using these repositories to integrate ...
11 months ago Darkreading.com
Misconfigured Firebase Instances Expose 125 Million User Records - Hundreds of websites misconfigured Google Firebase, leaking more than 125 million user records, including plaintext passwords, security researchers warn. It all started with the hacking of Chattr, the AI hiring system that serves multiple ...
8 months ago Securityweek.com
New "MITRE ATT&CK-like" framework outlines software supply chain attack TTPs - A new open framework seeks to outline a comprehensive and actionable way for businesses and security teams to understand attacker behaviors and techniques specifically impacting the software supply chain. The Open Software Supply Chain Attack ...
1 year ago Csoonline.com
Experts warn of DDoS attacks using linux printing vulnerability - Benjamin Harris, CEO of cybersecurity firm watchTowr, said CUPS itself is the software that handles printing on a Linux machine, and thus, by virtue of Linux not being a 'desktop' operating system in many environments in the same way that ...
1 month ago Therecord.media

Latest Cyber News


Cyber Trends (last 7 days)


Trending Cyber News (last 7 days)