In the perfectly productive world, humans would be accounted as worthless, certainly in terms of productivity but also in terms of our feeble humanity. Unless we jettison this perfectionist attitude towards life that positions productivity and “material growth” above sustainability and individual happiness, AI research could be another chain in the history of self-defeating human inventions. Already we are witnessing discrimination in algorithmic calculations. Recently, a popular South Korean chatbot named Lee Luda was taken offline. “She” was modelled after the persona of a 20-year-old female university student and was removed from Facebook messenger after using hate speech towards LGBT people. Meanwhile, automated weapons programmed to kill are carrying maxims such as “productivity” and “efficiency” into battle. As a result, war has become more sustainable. The proliferation of drone warfare is a very vivid example of these new forms of conflict. They create a virtual reality that is almost absent from our grasp. But it would be comical to depict AI as an inevitable Orwellian nightmare of an army of super-intelligent “Terminators” whose mission is to erase the human race.
Robots and intelligent technology can now optimise something we’ve never been able to before: the bandwidth of employees. This has become increasingly more critical as staff adjust to remote working. By onboarding these new tools and incorporating them into the workforce, businesses can empower their staff to do more. They can automate mundane and repetitive tasks extremely quickly, giving their human colleagues more time to take on problem-solving and time-consuming tasks. In fact, 4 in 5 employees that use robots and digital workers say they have been beneficial with efficiency and collaboration, and are useful in easing the burden of administrative tasks. Employees have found that a ‘robotic helping hand’ has been most appreciated for sorting data and documents, providing prompts for pending tasks, and digitising paperwork. What’s also clear is that some businesses do have the right tools in place to help. In fact, half of UK employees said processes helped them do their job faster and collaborate better, both critical during the pandemic. However, for business leaders, the pressure to get automation right is huge. It’s a major investment of time, money, and energy for everyone involved.
Many enterprises are finding it difficult to scale beyond a few software robots or bots because they are automating a bad process that cannot scale. “Most businesses are automating processes through RPA and hyperautomation without first fully understanding their data and processes,” explained Gero Decker, CEO of Signavio, a SAP spinoff focused on business transformation. As enterprises pursue increased efficiencies, there is debate about whether it makes more sense to automate what exists or to fix it first. Automating a bad process may make it faster, but it may also suffer from chokepoints caused by integration with legacy systems or approval processes. Process mining can help a company fix a bad process first. Chris Nicholson, CEO of Pathmind, a company applying AI to industrial operations, argues, “The main challenge to overcome before applying process automation is to standardize the current processes performed by people. If they are not standardized, there can be no automation.” With process mining, companies can see whether their current processes are standardized so they know which problem they have to solve first: standardization or automation.
The cybersecurity industry has long been aware that sophisticated and well-funded actors were theoretically capable of advanced techniques, patience, and operating below the radar, but this incident has proven that it isn’t just theoretical. We believe the Solorigate incident has proven the benefit of the industry working together to share information, strengthen defenses, and respond to attacks. Additionally, the attacks have reinforced two key points that the industry has been advocating for a while now—defense-in-depth protections and embracing a zero trust mindset. Defense-in-depth protections and best practices are really important because each layer of defense provides an extra opportunity to detect an attack and take action before they get closer to valuable assets. We saw this ourselves in our internal investigation, where we found evidence of attempted activities that were thwarted by defense-in-depth protections. So, we again want to reiterate the value of industry best practices such as outlined here, and implementing Privileged Access Workstations (PAW) as part of a strategy to protect privileged accounts.
AI transformation touches all aspects of the modern enterprise including both commercial and operational activities. Tech giants are integrating AI into their processes and products. For example, Google is calling itself an “AI-first” organization. Besides tech giants, IDC estimates that at least 90% of new organizations will insert AI technology into their processes and products by 2025. ... First few projects should create measurable business value while being attainable. This is important for the transformation to gain trust across the organization with achieved projects and it creates momentum that will lead to AI projects with greater success. These projects can rely on AI/ML powered tools in the marketplace or for more custom solutions, your company can run a data science competition and rely on the wisdom of hundreds of data scientists. These competitions use encrypted data and provide a low cost way to find high performing data science solutions. bitgrit is a company that helps companies identify AI use cases and run data science competitions. Implementing process mining tools is one of those easy-to-achieve and impactful projects. For example, QPR’s Process Analyzer tool has an extensive set of ready-to-use process mining analyses, including ready-to-use clustering analysis and process predictions, as well as a platform for machine learning based analyses.
Machine-learning researchers are focused on attacks that pollute machine learning data, epitomized by presenting two seemingly-identical image of, say, a tabby cat, and having the AI algorithm identify it as two completely different things, he said. More than 2,000 papers have been written in the last few years, citing these sorts of examples and proposing defenses, he said. "Meanwhile, security professionals are dealing with things like SolarWinds, software updates and SSL patches, phishing and education, ransomware, and cloud credentials that you just checked into Github," Anderson said. "And they are left to wonder what the recognition of a tabby cat has to do with the problems they are dealing with today." ... Anderson shared a red team exercise conducted by Microsoft where the team aimed to abuse a Web portal used for software resource requests and the internal machine-learning algorithm that determines automatically to which physical hardware it assigns a requested container or virtual machine. The red team started with credentials for the service, under the assumption that attackers will be able to gather valid credentials - either by phishing or because an employee reuses their user name and password.
In its Thursday blog, the Microsoft team says the compromise techniques leveraged by the SolarWinds hackers included "password spraying, spear-phishing and use of webshell through a web server and delegated credentials." Earlier this week, acting CISA Director Brandon Wales told The Wall Street Journal that the SolarWinds cyberespionage operation gained access to targets using a multitude of methods, including password spraying and through exploits of vulnerabilities in cloud software (see: SolarWinds Hackers Cast a Wide Net). "As part of the investigative team working with FireEye, we were able to analyze the attacker’s behavior with a forensic investigation and identify unusual technical indicators that would not be associated with normal user interactions. We then used our telemetry to search for those indicators and identify organizations where credentials had likely been compromised by the [SolarWinds hackers]," Microsoft's security team says. But Microsoft says it's found no evidence that the SolarWinds hackers used Office 365 as an attack vector. "We have investigated thoroughly and have found no evidence they [SolarWinds] were attacked via Office 365," the Microsoft researchers say. "The wording of the SolarWinds 8K filing was unfortunately ambiguous, leading to erroneous interpretation and speculation, which is not supported by the results of our investigation."
For many, a distributed hybrid workforce is the new normal, vastly expanding their threat landscape and making it more challenging to secure data and IT infrastructure. In this environment, companies need to pivot their defensive capacity, ensuring that they are prepared to meet the moment (i.e., the threats). When considering cybersecurity threats, we often think of shady cybercriminals or nation-states hacking company networks. After all, when these incidents occur, they make worldwide news headlines. For most companies, however, external bad actors aren’t the most critical risk. A company’s employees often pose a more prominent and – luckily – a more manageable cybersecurity threat. IBM estimates that human error causes nearly a quarter of all data breaches. Additionally, employees commonly and inadvertently compromise company data through poor password hygiene, accidental data sharing, improper technology use, phishing scams, and more. Some employees will also act maliciously, intentionally stealing company data for profit, retribution, or fun. The market for sensitive data is so prolific that some cybersecurity experts predict the emergence of insiders-as-a-service as bad actors capitalize on remote work trends to infiltrate companies.
In Public Safety arena using biased data to train the AI to identify criminals using cyber forensics can lead to the wrongful conviction of innocent people as the output of the software was influenced by racial and ethnicity data points introduced as either the code used was not tested properly or used wrong data sets for testing resulting in destroying lives. Apart from the bias in the data set we have also seen that during any application or transactional data processing there is no transparency as to find out why this decision was taken, which parameter influenced it and why did the algorithm took additional steps to mitigate it? All these can be easily answered by embedding explainability and transparency in the AI design processes to provide understandability of the context and interpretability of the decision by AI. Thus we need Responsible AI which is the practice of using AI with good intention to empower employees and businesses, and fairly impact customers and society – allowing companies to engender trust and scale AI with confidence along with the purpose of providing a framework to ensure the ethical, transparent and accountable use of AI technologies consistent with user expectations, organizational values and societal laws and norms.
Many companies struggle with defining an incident. To us, an incident is when a service or feature functionality is degraded. But defining "degraded" contains a multitude of possibilities. One could say "degraded" is when something isn’t working as expected. But what if it’s better than expected? What’s the expected behavior? Do you define it based on customer impact? Do you wait until there’s customer impact to declare an issue an incident? This is where having a common and shared understanding of the normal operating behavior of the system and formalizing these in feature/service level objectives and indicators are key. We have to know what we expect, to know when a degradation becomes an incident. But, defining service level objectives for legacy services already in operation takes a significant investment of time and energy that might not be available right now. That’s the reality in which we frequently operate, trading off efficiency with thoroughness, as Hollnagel (2009) points out. We handle this tradeoff with a governing set of generic thresholds to fill in for services without clear indicators. At Twilio we have a lot of products, running the gamut from voice calls, video conferencing, and text messages, to email and two factor authentication.
Quote for the day:
"Don't look back. Something might be gaining on you." -- Satchel Paige