• Search for:
    22Dec, 2023
    Indian tech giant HCL investigating ransomware attack

    Indian information technology company HCL Technologies reported a ransomware attack to regulators on Wednesday and said that it is investigating the incident.

    In a filing with the National Stock Exchange of India, the company said it “has become aware of a ransomware incident in an isolated cloud environment for one of its projects.”

    “There has been no impact observed due to this incident on the overall HCLTech network. Cybersecurity and data protection is a top priority for HCLTech,” the company’s secretary Manish Anand said.

    “A detailed investigation is underway in consultation with relevant stakeholders to assess the root cause and take remedial action as necessary. This is for your information and records.”

    HCL Technologies, based in Noida, is one of the largest tech companies in the world with more than 225,000 employees spread across 52 countries. The company reported revenues of $13 billion for fiscal 2023. HCL Technologies shares fell 3.24% on Wednesday.

    This is the latest ransomware attack in 2023 to target a major Indian corporation. The country’s largest drugmaker, Sun Pharmaceuticals, confirmed a ransomware attack in regulatory filings in March, warning that the incident involved the theft of company data and personal information.

    Tata Power reported a cyberattack in October 2022 that affected its billion-dollar power business. Tata did not call the incident a ransomware attack but reported that it was forced to “restore” systems and segment the affected networks in order to protect other parts of their business.

    In April, India’s Computer Emergency Response Team (CERT-In) said that reported ransomware attacks on organizations in the country increased 53% throughout 2022.

    “IT & [Information Technology Enabled Services] was majorly impacted sector followed by Finance and Manufacturing. Ransomware players targeted critical infrastructure organizations and disrupted critical services in order to pressurize and extract ransom payments,” they said.

    Resource : https://therecord.media/hcl-india-ransomware-attack?&web_view=true

    6Nov, 2023
    Role of AI in management education

    With the rapidly accelerating use of artificial intelligence and machine learning across segments, b-schools are in the process of updating their curricula in these areas. They are also integrating AI into their teaching methodology.

    Sangeeta Shah Bhardwaj, professor (information management) at MDI Gurgaon, says that though there may be business leaders who are apprehensive about AI, they would have no choice but to adopt it when others start using AI-powered tools. MDI, which has a programme in business analytics, has also introduced courses like geospatial AI and AI in marketing. The geospatial AI course looks at the impact location data is having on organisations and in the startup ecosystem. It will introduce to students the principles for assessing or creating location AI-based products and interfaces.

    “AI is being built as part of the process of learning management systems,” says Bhardwaj. “For instance, if there is an online group discussion, AI-enabled LMS will evaluate how students performed and will bring out traits like aggressiveness among other behaviour. This can help a professor in engaging students better.” She adds that students can also get personalised feedback regarding their career path with the help of AI.

    However, while AI can be a good supplement, especially in quantifiable business processes such as finance, operations, supply chain and sales, it may not solve the complexities our business leaders face. “A few months ago, an American lawyer used AI to prepare a case,” says Vivek Suneja, dean, FMS Delhi. “When the case went to court, it had a lot of incorrect information. AI manufactured information; basically it falsified data.”

    Moreover, there are limitations to the use of AI in human behavioural processes and decisions such as people engagement, succession planning, leadership and values and culture. Human behaviour is not easy to predict. AI will keep developing and become more sophisticated, but it may not be able to replace human traits, says Vishal Talwar, director, IMT Ghaziabad. “Nowadays, the students are more evolved in using such technologies and how management institutions are able to harness and ride that particular wave is something that we have to decide,” he says.

    Business leaders will also have to prioritise technology so that investments are made in the areas where the impact is on the quality of decision-making. “We have to look at robots and human beings working together,” says Ranjan Banerjee, dean, BITSoM, Mumbai. “Robots handle programmable aspects, leaving judgment, empathy, insight and relational aspects to humans. Management education will view technology not as a business function but as a horizontal, embedding technology in function.”

    27Oct, 2023
    13 AI tools that made it to the list of Time Magazine’s Best Inventions of 2023

    It is that time of the year again. Time Magazine has put together the best 200 inventions for the year 2023. The announcement includes ingenious inventions from across domains such as accessibility, Artificial Intelligence, apps and software, AR/VR, beauty, consumer electronics, etc. The list features an assortment of gadgets and software from established technology brands as well as barely known startups.

     AI tools have been making waves across the world since the launch of OpenAI’s ChatGPT. Ever since then, there has been a barrage of AI tools and applications. Time Magazine’s list included some of the best AI apps and tools that caught our attention. Here they are:

    GPT-4 from OpenAI

    The coveted magazine has described GPT-4 as the most powerful AI model that is accessible to the public. GPT-4 excels at verbal reasoning and can explain complex concepts in simple language. In September, the makers rolled out the ability to let users interact with the model using voice and images as inputs. Later, in a subsequent update, GPT-4V verbally describes the contents of a picture in natural language. GPT-4 is currently accessible only to ChatGPT Plus users, however, there are numerous ways to use GPT-4 for free with tools like Bing, HuggingFace, Poe, etc. 

    Runway Gen-2
    Gen-2 is an AI video software created by Runway, a Google-backed startup. It is to be noted that visual effects tools created by this company were used in the Hollywood blockbuster ‘Everything Everywhere All at Once’. Gen-2 is capable of creating videos from text prompts, images, or other videos as prompts. We tried the Gen-2 tool and were impressed by the output. Although not perfect, there is room for refinement such as offering precise descriptions or using images as prompts. 

    Photo Editing Outside The Box

    Earlier this year, Adobe endowed Photoshop with its AI image generator – Firefly. The new addition has brought features such as Generative Expand and Generative Fill. While Generative Expand allows users to fill in content well beyond the borders of an image, Generative Fill lets users add or delete aspects of an existing image using text prompts. One need not have technical expertise to use this. The biggest advantage of this update is that it amplifies a user’s imagination.

    Humane’s AI Pin

    Ai Pin is the first product from Humane, an AI startup that has been founded by former execs of Apple. The lightweight device can be pinned to your garment and projects calls and other information straight to your palms. This essentially makes smartphones obsolete. While the company has yet to launch the device officially, it turned heads during the recent Paris Fashion Week when it was spotted on supermodel Naomi Campbell. AI Pin is made using proprietary software and OpenAI’s GPT. It allows users to ask complex questions, make calls, and send and receive text messages. There is also a built-in camera that can identify objects and give information. The device has been developed keeping in mind the privacy and security of the users. It is likely to be launched on November 9.

    Dall-E 3 from OpenAI

    Dall-E 3 is OpenAI’s ultimate AI tool for making images using text prompts. However, the company has enhanced the text-to-image model significantly. The model which was invented by OpenAI’s Aditya Ramesh can now be accessed by ChatGPT Plus and Enterprise users. The AI tool enables anyone to create images and they do not need to learn any technical language.

    Meta SeamlessM4T

    In August, Meta released SeamlessM4T, an AI model that can translate and transcribe nearly 100 languages. SeamlessM4T is the first all-in-one multilingual multimodal AI translation and transcription model. It is capable of translating in a variety of formats such as text-to-speech, speech-to-speech, speech-to-text, or text-to-speech. Despite working with more languages, the system is claimed to be more efficient and is less-error prone. Besides, the model is open source, which allows researchers around the world to use its code. 

    Dedrone City-Wide Drone Detection

    Dedrone is a leading provider of airspace security. The company has developed Dedrone City-Wide which is a drone detection solution that gives security and a quick view of the airspace with their app. According to Time Magazine, the device offers a virtual shield around a geographical area and warns law enforcement agencies within seconds in case a drone enters the designated airspace.

    Alitheon FuturePrint

    This is one of the handiest tools that can help you determine the authenticity of anything – from jewellery to handbags. FuturePrint from Alitheon is essentially an optical AI technology that can authenticate products by simply taking pictures. The software examines intricate details on the surface of objects and fingerprints them into something known as unique mathematical identities. 

    Astounding Mimicry: So-VITS-SVC

    The open-source software So-VITS-SVC allows users to train a neural network based on a singer’s voice and later produce songs using the AI-generated voice. The tool came to the limelight after the AI-generated track ‘Heart on my sleeve’ with convincing vocals of Drake and the Weeknd went viral. This has paved the way for many AI tools to create voice-based services.

    Stopping Wildfires: AlertCalifornia and Cal Fire AI Wildfire Detector

    This is an AI fire detection tool that can come as a saviour, as detecting wildfires is often a daunting task. Cal Fire is an outcome of the efforts by the University of California San Diego’s public safety program – AlertCalifornia. The program trained AI to detect smoke and other early signs of fire with the help of over 1,050 cameras deployed across the forest. According to the listing in Time, the AI system correctly identified 77 fires much before 911 calls. 

    Trailguard AI 

    Resolve, a US-based environmental organisation, is offering an innovative way to counter the threat of poaching that is driving many species of animals to extinction. The company has developed a tool named Trailguard AI that has tiny cameras that are powered by Intel that monitor endangered species and spot poachers. 

    Project Gutenberg Open Audiobook Collection

    After becoming the first free provider of ebooks, Project Gutenberg is now opening the world to its audiobook collection. Microsoft and MIT have collaborated to make the Open Audiobook Collection with the help of text-to-speech technology to turn close to 5,000 books into audiobooks. These are now available on Spotify and the software that is powering this is also being offered for free. 

    Audioshake

    Audioshake is a company headed by Jessica Powell. The company came to prominence owing to its use of AI to deconstruct songs into instrumentals. According to Powell, the company makes audio interactive and editable and this makes it possible to fuel some of the most practical uses for artists to earn money.

    The above-listed AI tools come with a plethora of use cases, tell us what tool are you excited about in the comments section. 

    Resource : https://indianexpress.com/article/technology/artificial-intelligence/time-magazine-best-inventions-of-2023-top-ai-tools-9000899/

    10Oct, 2023
    Latest Balada Injector Campaign Targets Unpatched tagDiv Plugin

    A  new wave of Balada malware injection attacks has been found exploiting a vulnerable tagDiv premium theme plugin to target Newspaper and Newsmag websites. The flaw in the question is an unauthenticated XSS vulnerability in the plugin that was first disclosed in September.

    The plugin is used by over 135,000 users, which emphasizes the risk associated with the attack.

    Attack methods

    The attackers are using different tactics and techniques to stay under the radar while duping users into visiting fake websites.

    • The first wave of attacks was launched by injecting two variants of Balada injector into public WordPress pages. 
    • While the first variant was detected on over 4,000 sites, the second variant was found on another 1,000 websites.
    • In the second wave of attacks, the attackers created malicious admin usernames and email IDs for the targeted sites to initiate the infection process or to plant backdoors.
    • The third wave involved planting the malware injector in the Newspaper theme’s 404.php file. 
    • Around September 17–18, in the fourth wave, the attackers shifted the existing infection process and began using a malicious wp-zexit plugin installation that mimicked the original installation page.
    • In the fifth wave, that started on September 21, attackers changed the location of the injection to the std_live_css_local_storage option in the WordPress database. Additionally, they registered three new domains within a period of seven seconds.  
    • The sixth wave began on September 29 and involved several different scripts that loaded malware from the subdomains of promsmotion[.]com.

    A similar incident in the past

    This is not the first time that malware operators have leveraged the plugin to target websites. The malware injector was part of a massive campaign that infected over one million WordPress websites over a duration of around five years, starting in 2017. The subdomains of these WordPress sites were injected with malicious scripts that redirected visitors to scam sites including fake tech support, fraudulent lottery wins, and push notification scams.

    Ending notes

    Researchers have shared the list of malicious domains and IP addresses to help organizations detect such threats. Updating the plugin to the latest version 4.2 and using website scanners helps users prevent infection. Furthermore, it is recommended to remove all unwanted admin users and redundant plugins to stay safe.

    Resource : https://cyware.com/news/latest-balada-injector-campaign-targets-unpatched-tagdiv-plugin-37f55318

    13Jul, 2023
    NASA Engineers Help Create A Virtual World of Data

    For almost as long as it’s been a concept, NASA has been on the cutting edge of virtual reality (VR) technology. However, the space has seen a renaissance since the bulky headsets of the 1990s. Several high-profile companies now use VR for immersive video games and virtual chat rooms, but to some, this technology has a use beyond entertainment.

    NASA’s Jet Propulsion Laboratory in Southern California has been trying to improve data visualization for decades. When pictures come back from an alien world, scientists usually survey the environment with a panoramic photo, but this method has limitations. When observing a flat image, judging distances can be difficult.

    “Looking at pictures on a screen is a much different experience to walking through a canyon,” said Scott Davidoff, manager of the Human Centered Design Group at JPL.

    Davidoff began experimenting with VR using image data from the Curiosity rover on Mars. There had been attempts to solve the visual distance issue, like using 3D imagery viewable with red and blue filtered glasses, but nothing made scientists feel like they were really there. Davidoff and his colleagues decided to wrap the panorama around the scientist in a virtual environment. Geologists using VR in this way reported feeling like they were in the real Martian environment. They could ascertain the distance and size of features much more quickly and accurately than they could with a flat display, which sparked another idea. Davidoff thought that if viewing imagery in VR helped, it could be a game changer for more complex “multi-dimensional” data.

    “When you look at a network diagram as a system in 3D, it turns out your perception does something different,” Davidoff said. “We made a data world where an analyst could look at any science or engineering problem and see patterns and correlations more clearly than they can in a flat version.”

    At the same time, Ciro Donalek and George Djorgovski at the California Institute of Technology (Caltech) were researching how to use immersive environments for scientific data visualization and collaboration.

    The three of them teamed up to create software that uses 3D visualizations to examine relationships between points of data. Much like how viewing a three-dimensional Martian panorama allows for better interpretation, seeing data in virtual space makes spotting correlations easier.

    Obtaining an exclusive license from Caltech, which manages JPL, Donalek, Djorgovski, and new CEO Michael Amori founded Virtualitics Inc. of Pasadena, California, in 2016, with Davidoff in an advisory role. To further build out the software, they’ve incorporated features such as artificial intelligence to point out patterns and relationships in the visualized data.

    “We call it intelligent exploration. You’re using AI and 3D visualizations to quickly identify drivers and relationships in your data, and drive understanding in ways 2D graphs aren’t capable of,” Donalek, now CTO of Virtualitics, said. “You can start getting insights from your data right away because it’s literally drag and drop.”

    The software works on both desktop and VR and is compatible with several popular headsets. While the platform is designed for use anywhere there’s a significant amount of data, it is currently most often used in banking, retail, and medical research. It can visualize anything from a spreadsheet to a large “data lake” without having to move the data from where it resides. Donalek says the company maintains a relationship with JPL, and many of its 60 employees come from the center or Caltech.

    “One of them is from MIT, but we don’t hold it against him,” Donalek joked.

    Whether analysts are looking at science from the surface of Mars or stock trades on Wall Street, NASA’s visualization efforts have led to a whole new dimension of research.

    NASA has a long history of transferring technology to the private sector. The agency’s Spinoff publication profiles NASA technologies that have transformed into commercial products and services, demonstrating the broader benefits of America’s investment in its space program. Spinoff is a publication of the Technology Transfer program in NASA’s Space Technology Mission Directorate (STMD).

    Resource : https://www.nasa.gov/directorates/spacetech/spinoff/NASA_Engineers_Help_Create_A_Virtual_World_of_Data

    29May, 2023
    How to turn on Google’s new AI Search Generative Experience

    For some users, Google has completely overhauled Search with AI, pushing what it calls the “Search Generative Experience.” While the experience is limited, some users have the option to turn it on now. Here’s how to turn on Search Generative Experience – SGE – and Google with AI.

    It’s easy to assume most people are used to the Google Search experience we’ve known and appreciated for years on end. While the foundation stays the same, things are starting to shift a little bit, with a Google experience that is bolstered by smarter, more contextual, and more dynamic results.

    Using the progress Google has made with AI up to this point, the company is overhauling Search with several new AI features and tools under the Search Generative Experience umbrella. That includes new AI-powered snapshots, which showcase quick, contextual results for your query or prompt. Rather than rattling off links, Google Search will develop a summary in answer to the question utilizing multiple sources – shown off to the right.

    On top of that, the SGE is meant to facilitate conversations. In effect, this means you can keep asking follow-up questions and get more information as the questions come up. This feature is a breath of fresh air in comparison to standard AI-free Search, which requires additional and alternatively worded searches. Rather than starting over, simply ask for a quick follow-up and get information that’s relevant to the entire conversation.

    Since the program is in its baby stages, there are a couple of criteria that need to be met before users can turn the feature on in Search. Users need to have signed up for Labs and be a Google One Premium (2+ TB) subscriber. These limitations are temporary, but that’s how it stands at the time of writing. Even if you’re not a subscriber, sign up for the free Labs anyway because that restriction might fall eventually.

    We have a full guide on signing up for the SGE Lab, and it takes only a couple of minutes.

    Once you’re signed up for Labs, complete the following:

    1. In Google Search – mobile or web – look for the Labs icon. Click or tap it.
    2. Look for the SGE card. Tap or click the toggle so it’s set to “on.”
    3. Hit I agree.

    Once you do that, you have access to Google’s new AI Generative Search Experience. Our breakdown of the current state of the feature is a helpful guide to using it, though the best approach is to ask natural questions that you wouldn’t normally be able to Google with much success.

    If you’re not a Google One subscriber, don’t worry. The SGE feature will likely be on its way to others who have signed up for the Lab in preview, though it might take a little time.

    Resource : https://9to5google.com/2023/05/26/turn-on-google-search-ai/

    6Apr, 2023
    NASA’s Ingenuity Mars Helicopter breaks world records, again

    NASA’s Ingenuity Mars Helicopter has been flying on the red planet’s surface for over two years now, and it has covered a distance of over 11 kilometres in that time. While 11 kilometres in two years may not seem like much in Earth terms, Ingenuity is actually breaking world records on the Mars.

    Ingenuity’s 49th flight, which happened on Sunday, April 2, broke two different records–that for the highest altitude and the highest speed–achieved by any self-propelled aircraft outside of Earth. During its 49th flight, the little rotorcraft flew up to a height of 16 metres and reached a top speed of 23.4 kilometres per hour.

    The rotorcraft set its previous flight height record on the red planet on December 3, 2022, when it flew about 14 metres in the air. Its previous flight speed record of 19.8 kilometres per hour was achieved on three different occasions–December 10, June 11, and April 8.

    Its longest flight so far was its record-breaking 25th flight on April 8, 2022, when it covered a distance of a whopping 704 metres in what is still its longest flight and was its fastest flight until flight 49.

    What is Ingenuity?

    Ingenuity is a small solar-powered helicopter that landed on the Martian surface on February 18, 2021, along with the Perseverance Rover. On April 19, the same year, it created history by completing the first powered extraterrestrial flight in human history. During that flight, it hovered and landed and the same spot for a flight that lasted 39.1 seconds, creating history.

    Why are Ingenuity’s records so impressive?

    While these distances may seem small for us here living on Earth, do remember that Mars is more than 225 million kilometres away right now. The distance between the two planets means that it takes between 5 to 20 minutes for a signal to travel from one to another.

    Adding to this signal delay, Ingenuity also has to survive the harsh conditions on Mars. The low atmospheric density on the planet means that Ingenuity must work much harder than helicopters here on Earth in order to be able to fly. Also, it has had to survive “continent-sized” dust storms in the past and other Martian hazards in the past.

    Resource : https://indianexpress.com/article/technology/science/nasa-ingenuity-mars-helicopter-8539782/

    14Mar, 2023
    NASA To Measure Forest Health from Above

    In places across the U.S., tree cover is shrinking – forests are burned by wildfires on the West Coast and drowned by rising sea levels along the East. From the ground, it’s hard to assess the scale of the losses and the effects disappearing trees have on atmospheric carbon dioxide levels and climate change.

    NASA research scientist Jon Ranson is working to improve new technologies for studying trees from above, so future Earth-observing missions can more accurately assess forest health.

    “Trees do a huge service for the planet, in terms of sequestering carbon dioxide, taking it out of the atmosphere and putting it into wood,” Ranson said, “but trees are very sensitive to our changing climate. We’re trying to see the changes going on in forest ecosystems. If you detect things early enough, you might be able to do something about it.”

    When measured from aircraft and satellites, the wavelengths of light reflected by plants tell scientists about the amount of photosynthesis occurring, and therefore how much atmospheric carbon dioxide trees take up and store. The current standard for studying vegetation is called NDVI, or Normalized Differential Vegetation Index, which is the average of two broad portions of the infrared spectrum. NASA’s NDVI record goes back 40 years, providing a low-resolution but accurate picture of forest health.

    While NDVI is good at assessing vegetation quantity and vigor broadly, Ranson said, breaking down infrared and visible light into many more wavelengths, a technology called hyperspectral imaging, can provide insight on plants’ water content, chlorophyll and even changes in health. 

    “Vegetation has these broad spectral properties,” said Ian Adams, Earth Sciences Division Technologist at NASA’s Goddard Space Flight Center. “With hyperspectral imaging, you get lots of different measurements at smaller, closer together frequencies. There is a lot more information we can pull out if we can get better spectral resolution.”

    By dramatically increasing the number of frequencies available for researchers to study, Ranson’s work serves the priorities of the National Academies of Science’s most recent Earth Science decadal survey, which determines the field’s long-term priorities. The survey lists “surface topography and vegetation,” including forests, as a key area of study in need of more advanced technology.

     “Strategically for NASA, and more broadly for the remote sensing community, hyperspectral is one of the areas we see as the future,” Adams said.

    To make use of hyperspectral imaging by future orbital missions, data analysis techniques are first proven closer to the ground.

    Ranson’s team fitted a Skyfish drone (UAV) from partner institution Virginia Tech with a visible and infrared (VIS/IR) hyperspectral camera and lidar equipment. They flew the imaging equipment over forests near Blacksburg, VA, in a region called Mountain Lake. Comparing their UAV observations with actual CO2 levels recorded by sensors on a nearby National Ecological Observatory Network tower, Ranson’s team was able to refine calculations about how much carbon the forest removed from the atmosphere.

    These comparisons allowed them to further refine techniques for interpreting the hyperspectral data. For instance, Plants stressed by too much sunlight may release pigments to shield their chloroplasts, a condition his sensors can detect. If plants get too much shade, they may grow leaves with larger surface-areas, which can cause the sensor to overestimate plant productivity. Ranson wants to add short-wave infrared sensors to better distinguish reflections from leaves and other parts of plants, eliminating another possible source of error.

    The view from space: Seeing the forest for the trees

    Ranson’s goal is to study forests on a global scale from space.

    “We’re trying to get the complete picture,” Ranson said. “So that when we go to space, we know how to find the answer to the question, ‘Are our forests healthy? And if not, why not?’”

    Scientists and engineers at Goddard have been developing a mission concept called Concurrent Artificially intelligent Spectrometry and Adaptive Lidar System, or CASALS. If selected, CASALS would send a satellite with lidar and hyperspectral cameras into space. A constellation of such satellites could take the frequent measurements necessary to assess changes in forest productivity over time, using models perfected by Ranson’s UAV flights.“Forests take up as much carbon as the ocean does, yet forests cover only 9 percent of Earth’s surface,” said Ranson. “So, if something goes wrong with our forests, it’s a dramatic issue. We’ve got this great, great resource in forests, and we need to care for them so they’ll care for us.”

    Resource : https://www.nasa.gov/feature/nasa-to-measure-forest-health-from-above

    18Feb, 2023
    GitHub Copilot update stops AI model from revealing secrets

    GitHub has updated the AI model of Copilot, a programming assistant that generates real-time source code and function recommendations in Visual Studio, and says it’s now safer and more powerful.

    The company says the new AI model, which will be rolled out to users this week, offers better quality suggestions in a shorter time, further improving the efficiency of software developers using it by increasing the acceptance rate.

    Copilot will introduce a new paradigm called “Fill-In-the-Middle,” which uses a library of known code suffixes and leaves a gap for the AI tool to fill, achieving better relevance and coherence with the rest of the project’s code.

    Additionally, GitHub has updated the client of Copilot to reduce unwanted suggestions by 4.5% for improved overall code acceptance rates.

    “When we first launched GitHub Copilot for Individuals in June 2022, more than 27% of developers’ code files on average were generated by GitHub Copilot,” Senior Director of Product Management Shuyin Zhao said.

    “Today, GitHub Copilot is behind an average of 46% of a developers’ code across all programming languages—and in Java, that number jumps to 61%.”

    More secure suggestions

    One of the highlight improvements in this Copilot update is the introduction of a new security vulnerability filtering system that will help identify and block insecure suggestions such as hardcoded credentials, path injections, and SQL injections.

    “The new system leverages LLMs (large language models) to approximate the behavior of static analysis tools—and since GitHub Copilot runs advanced AI models on powerful compute resources, it’s incredibly fast and can even detect vulnerable patterns in incomplete fragments of code,” Zhao said.

    “This means insecure coding patterns are quickly blocked and replaced by alternative suggestions.”

    The software company says Copilot may generate secrets like keys, credentials, and passwords seen in the training data on novel strings. However, these aren’t usable as they’re entirely fictitious and will be blocked by the new filtering system.

    The appearance of these secrets in Copilot’s code suggestions has caused fierce criticism from the software developing community, with many accusing Microsoft of using large sets of publicly available data to train its AI models with little regard to security, even including sets that mistakenly contain secrets.

    By blocking unsafe suggestions in the editor in real-time, GitHub might also provide some resistance against poisoned dataset attacks aiming to covertly train AI assistants to make suggestions containing malicious payloads.

    At this time, Copilot’s LLMs are still being trained to distinguish between vulnerable and non-vulnerable code patterns, so the AI model’s performance on that front is expected to improve gradually in the near future.

    Resource : https://www.bleepingcomputer.com/news/security/github-copilot-update-stops-ai-model-from-revealing-secrets/

    14Feb, 2023
    Skyview Networks Suffers Security Incident

    An unauthorized person or group gained access to internal systems used by Skyview Networks this week, disrupting the delivery of the CBS World News Roundup and other programming to radio affiliates on Monday.

    The issue was confirmed in an email sent to Radio Ink by Steve Jones, the president and CEO of Skyview, who said the company “experienced unauthorized access to its IT environment, resulting in interrupted technical services.”

    “Skyview activated its comprehensive incident response plan,” Jones affirmed. “We have initiated advanced monitoring and threat detection across our systems and data. We have also involved industry-leading forensic experts to help us contain and investigate the incident. These efforts are fully underway.”

    Jones said specific information about the cyberattack could not be disclosed at this time, because the investigation and response are still ongoing. “Our highest priority is the security and protection of all data, and we are cooperating with authorities on this matter,” he affirmed.

    Stations were notified about the issue Monday morning when CBS News Radio sent a bulletin urging them to replace the slot where CBS World News Round would normally air with local programming, according to a copy of the message obtained by Radio Ink. Some stations were eventually able to obtain CBS News Radio spots, including the hourly bulletins, through other means while Skyview worked on the issue.

    The problem lasted well into the evening, with stations unable to get access to some of Skyview’s other programming, including XYZ with Erik Zachary. Web portals used by stations to access spots and programming were down throughout the day, and equipment used by radio affiliates to obtain Skyview content experienced severe problems when reset, according to numerous station officials.

    In a memo sent to affiliates on Tuesday and obtained by Radio Ink, Skyview warned stations not to reboot their XDS receivers while the company continues to investigate the issue.

    Jones said Skyview apologized for the disruption caused to client stations, and that the company will “provide our business clients with service updates and request that, as you receive this information, you treat it with sensitivity as we pursue a successful resolution.”

    Resource : https://radioink.com/2023/02/01/skyview-networks-suffers-security-incident/