Each January on the SEI Weblog, we current the ten most-visited posts of the earlier 12 months. This 12 months’s high 10 record highlights the SEI’s work in software program acquisition, synthetic intelligence, giant language fashions, safe coding, insider danger mitigation, and enterprise danger administration. The posts, which had been printed between January 1, 2024, and December 31, 2024, are introduced under in reverse order based mostly on the variety of visits.
by Ipek Ozkaya and Brigid O’Hearn
The fiscal 12 months 2022 Nationwide Protection Authorization Act (NDAA) Part 835, “Impartial Research on Technical Debt in Software program-Intensive Methods,” required the Secretary of Protection to have interaction a federally funded analysis and growth heart (FFRDC) “to check technical debt in software-intensive techniques.” To fulfill this requirement and lead this work, the Division of Protection (DoD) chosen the Carnegie Mellon College (CMU) Software program Engineering Institute (SEI), which is a acknowledged chief within the follow of managing technical debt. In line with NDAA Part 835, the aim of the examine was to offer, amongst different issues, analyses and suggestions on quantitative measures for assessing technical debt, present and greatest practices for measuring and managing technical debt and its related prices, and practices for decreasing technical debt.
Our workforce spent greater than a 12 months conducting the unbiased examine. The report we produced describes the conduct of the examine, summarizes the technical developments noticed, and presents the ensuing suggestions. On this SEI Weblog put up, we summarize a number of suggestions that apply to the DoD and different growth organizations looking for to research, handle, and cut back technical debt. You could find a whole dialogue of the examine methodology, findings, and suggestions within the SEI’s Report back to the Congressional Protection Committees on Nationwide Protection Authorization Act (NDAA) for Fiscal Yr 2022 Part 835 Impartial Research on Technical Debt in Software program-Intensive Methods.
Learn the put up in its entirety.
by Douglas Schmidt and John E. Robert
There may be appreciable curiosity in utilizing generative AI instruments, corresponding to giant language fashions (LLMs), to revolutionize industries and create new alternatives within the business and authorities domains. For a lot of Division of Protection (DoD) software program acquisition professionals, the promise of LLMs is interesting, however there’s additionally a deep-seated concern that LLMs don’t handle in the present day’s challenges as a result of privateness considerations, potential for inaccuracy within the output, and insecurity or uncertainty about methods to use LLMs successfully and responsibly. This weblog put up is the second in a sequence devoted to exploring how generative AI, significantly LLMs corresponding to ChatGPT, Claude, and Gemini, could be utilized throughout the DoD to reinforce software program acquisition actions.
Our first weblog put up on this sequence introduced 10 Advantages and 10 Challenges of Making use of LLMs to DoD Software program Acquisition and recommended particular use circumstances the place generative AI can present worth to software program acquisition actions. This second weblog put up expands on that dialogue by exhibiting particular examples of utilizing LLMs for software program acquisition within the context of a doc summarization experiment, in addition to codifying the teachings we discovered from this experiment and our associated work on making use of generative AI to software program engineering.
Learn the put up in its entirety.
by Robin Ruefle
Incident response is a important want all through authorities and trade as cyber menace actors look to compromise important belongings inside organizations with cascading, usually catastrophic, results. In 2021, for instance, a hacker allegedly accessed a Florida water therapy plant’s pc techniques and poisoned the water provide. Throughout the U.S. important nationwide infrastructure, 77 % of organizations have seen an increase in insider-driven cyber threats during the last three years. The 2023 IBM Value of a Knowledge Breach report highlights the essential function of getting a well-tested incident response plan. Corporations with out a examined plan in place will face 82 % larger prices within the occasion of a cyber assault, in contrast to people who have carried out and examined such a plan.
Researchers within the SEI CERT Division compiled 10 classes discovered from our greater than 35 years of growing and dealing with incident response and safety groups all through the globe. These classes are related to incident response groups contending with an ever-evolving cyber menace panorama. In honor of the CERT Division (additionally referred to the CERT Coordination Middle in our work with the Discussion board of Incident Response and Safety Groups) celebrating 35 years of operation, on this weblog put up we have a look again at a few of the classes discovered from our Cyber Safety Incident Response Staff (CSIRT) capability constructing experiences that additionally apply to different areas of safety operations.
Learn the put up in its entirety.
by Roger Black
In line with a 2023 Ponemon examine, the variety of reported insider danger incidents and the prices related to them continues to rise. With greater than 7,000 reported circumstances in 2023, the common insider danger incident price organizations over $600,000. To assist organizations assess their insider danger applications and determine potential vulnerabilities that would lead to insider threats, the SEI CERT Division has launched two instruments accessible for obtain on its web site. Beforehand accessible solely to licensed companions, the Insider Risk Vulnerability Evaluation (ITVA) and Insider Risk Program Analysis (ITPE) toolkits present sensible strategies to evaluate your group’s means to handle insider danger. This put up describes the aim and use of the toolkits, with a concentrate on the workbook elements of the toolkits which might be the first strategies of program evaluation.
Learn the put up in its entirety.
by David Svoboda
In current weeks a number of vulnerabilities have rocked the Rust neighborhood, inflicting many to query the protection of the borrow checker, or of Rust typically. On this put up, we study two such vulnerabilities: the primary is CVE-2024-3094, which includes some malicious information within the xz library, and the second is CVE-2024-24576, which includes command-injection vulnerabilities in Home windows. How did these vulnerabilities come up, how had been they found, and the way do they contain Rust? Extra importantly, may Rust be inclined to extra related vulnerabilities sooner or later?
Final 12 months we printed two weblog posts in regards to the safety offered by the Rust programming language. We mentioned the reminiscence security and concurrency security offered by Rust’s borrow checker. We additionally described a few of the limitations of Rust’s safety mannequin, corresponding to its restricted means to forestall varied injection assaults, and the unsafe key phrase, which permits builders to bypass Rust’s safety mannequin when needed. Again then, our conclusion was that no language may very well be absolutely safe, but the borrow checker did present important, albeit restricted, reminiscence and concurrency security when not bypassed with the unsafe key phrase. We additionally examined Rust by means of the lens of supply and binary evaluation, gauged its stability and maturity, and realized that the constraints and expectations for language maturity have slowly developed over the a long time. Rust is shifting within the course of maturity in the present day, which is distinct from what was thought of a mature programming language in 1980. Moreover, Rust has made some notable stability ensures, corresponding to promising to deprecate relatively than delete any crates in crates.io to keep away from repeating the Leftpad fiasco.
Learn the put up in its entirety.
by Ipek Ozkaya, Douglas Schmidt, and Michael Hilton
The preliminary surge of pleasure and worry surrounding generative synthetic intelligence (AI) is regularly evolving right into a extra life like perspective. Whereas the jury continues to be out on the precise return on funding and tangible enhancements from generative AI, the fast tempo of change is difficult software program engineering training and curricula. Educators have needed to adapt to the continued developments in generative AI to offer a practical perspective to their college students, balancing consciousness, wholesome skepticism, and curiosity.
In a current SEI webcast, researchers mentioned the influence of generative AI on software program engineering training. SEI and Carnegie Mellon College specialists spoke about using generative AI within the curriculum and the classroom, mentioned how school and college students can most successfully use generative AI, and regarded considerations about ethics and fairness when utilizing these instruments. The panelists took questions from the viewers and drew on their expertise as educators to talk to the important questions generative AI raises for software program engineering training.
This weblog put up options an edited transcript of responses from the unique webcast. Some questions and solutions have been rearranged and revised for readability.
Learn the put up in its entirety.
by Jeff Gennari, Shing-hon Lau, and Samuel J. Perl
Massive language fashions (LLMs) have proven a outstanding means to ingest, synthesize, and summarize data whereas concurrently demonstrating important limitations in finishing real-world duties. One notable area that presents each alternatives and dangers for leveraging LLMs is cybersecurity. LLMs might empower cybersecurity specialists to be extra environment friendly or efficient at stopping and stopping assaults. Nevertheless, adversaries might additionally use generative synthetic intelligence (AI) applied sciences in sort. Now we have already seen proof of actors utilizing LLMs to help in cyber intrusion actions (e.g., WormGPT, FraudGPT, and so forth.). Such misuse raises many vital cybersecurity-capability-related questions together with
- Can an LLM like GPT-4 write novel malware?
- Will LLMs change into important elements of large-scale cyber-attacks?
- Can we belief LLMs to offer cybersecurity specialists with dependable data?
The reply to those questions will depend on the analytic strategies chosen and the outcomes they supply. Sadly, present strategies and methods for evaluating the cybersecurity capabilities of LLMs are usually not complete. Not too long ago, a workforce of researchers within the SEI CERT Division labored with OpenAI to develop higher approaches for evaluating LLM cybersecurity capabilities. This SEI Weblog put up, excerpted from a lately printed paper that we coauthored with OpenAI researchers Joel Parish and Girish Sastry, summarizes 14 suggestions to assist assessors precisely consider LLM cybersecurity capabilities.
Learn the put up in its entirety.
by John E. Robert and Douglas Schmidt
Division of Protection (DoD) software program acquisition has lengthy been a fancy and document-heavy course of. Traditionally, many software program acquisition actions, corresponding to producing Requests for Data (RFIs), summarizing authorities rules, figuring out related business requirements, and drafting undertaking standing updates, have required appreciable human-intensive effort. Nevertheless, the appearance of generative synthetic intelligence (AI) instruments, together with giant language fashions (LLMs), presents a promising alternative to speed up and streamline sure elements of the software program acquisition course of.
Software program acquisition is considered one of many advanced mission-critical domains which will profit from making use of generative AI to reinforce and/or speed up human efforts. This weblog put up is the primary in a sequence devoted to exploring how generative AI, significantly LLMs like ChatGPT-4, can improve software program acquisition actions. On this put up we current 10 advantages and 10 challenges of making use of LLMs to the software program acquisition course of and counsel particular use circumstances the place generative AI can present worth. Our focus is on offering well timed data to software program acquisition professionals, together with protection software program builders, program managers, techniques engineers, cybersecurity analysts, and different key stakeholders, who function inside difficult constraints and prioritize safety and accuracy.
Learn the put up in its entirety.
by Mark Sherman
The common code pattern incorporates 6,000 defects per million traces of code, and the SEI’s analysis has discovered that 5 % of those defects change into vulnerabilities. This interprets to roughly 3 vulnerabilities per 10,000 traces of code. Can ChatGPT assist enhance this ratio? There was a lot hypothesis about how instruments constructed on high of huge language fashions (LLMs) may influence software program growth, extra particularly, how they may change the best way builders write code and consider it.
In March 2023 a workforce of CERT Safe Coding researchers—the workforce included Robert Schiela, David Svoboda, and myself—used ChatGPT 3.5 to look at the noncompliant software program code examples in our CERT Safe Coding customary, particularly the SEI CERT C Coding Normal. On this put up, I current our experiment and findings, which present that whereas ChatGPT 3.5 has promise, there are clear limitations.
Learn the put up in its entirety.
by Greg Touhill
The function of the chief data safety officer (CISO) has by no means been extra vital to organizational success. The current and near-future for CISOs might be marked by breathtaking technical advances, significantly these related to the inclusion of synthetic intelligence applied sciences being built-in into enterprise capabilities, in addition to emergent authorized and regulatory challenges. Continued advances in generative synthetic intelligence (AI) will speed up the proliferation of deepfakes designed to erode public belief in on-line data and public establishments. Moreover, these challenges might be amplified by an unstable world theater wherein nefarious actors and nation states chase alternatives to take advantage of any potential organizational weak point. Some forecasts have already characterised 2024 as a stress cooker atmosphere for CISOs. In such an atmosphere, expertise are important. On this put up I define the highest 10 expertise that CISOs want for 2024 and past. These suggestions draw upon my expertise because the director of the SEI’s CERT Division, in addition to my service as the primary federal chief data safety officer of the US, main cyber operations on the U.S. Division of Homeland Safety, and my prolonged navy service as a communications and our on-line world operations officer.
Learn the put up in its entirety.
Trying Forward in 2025
We publish a brand new put up on the SEI Weblog weekly. Within the coming months, search for posts highlighting the SEI’s work in synthetic intelligence, machine studying, cybersecurity, software program engineering, and extra.
