Jan 16, 2025

AI Asset Management Challenges Emerge Asset management challenges are coming to AI. As businesses start building out a catalogue of models, they will encounter challenges with size, portability, and discoverability. The industry will look for ways to get better compression with minimal reduction in accuracy from these assets so that they are more portable. There will be a need to effectively manage models, making them easy to find across organizations, and making them ever more interoperable. — Robert Elwell, VP of engineering, MacStadium Ransomware and Digital Extortion (R&DE) We expect R&DE incidents to continue at an elevated level in 2025, representing a significant threat to organizations of all sizes, industries, and geographies. 2024 was a record year for R&DE collectives with ZeroFox identifying an average of 388 incidents each month throughout 2024, compared to an average of 337 per month in 2023. Organizations that make up the manufacturing industry are likely to face the biggest threat from R&DE actors throughout 2025, with those within the retail, construction, healthcare, and technology sectors also prone to high levels of targeting. The greatest threat in early 2025 will very likely emanate from RansomHub, an extortion collective that was first observed in early 2024 and went on to become the most prominent R&DE outfit of the year. 2025 is likely to see an increasing number of new threat collectives, which continue to diversify the R&DE threat landscape. New collectives will also continue to develop and test new TTPs, such as increased emphasis on data extraction over traditional encryption methods, and to opt for double or triple extortion tactics in a bid to increase the chance of successful ransom demands. — Adam Darrah, VP of intelligence, ZeroFox Geopolitical & Cyber Convergence The cyber threat landscape in 2025 is expected to be heavily influenced by geopolitical developments, continuing the trend of increasing convergence between cyber and geopolitical spheres. Throughout 2024, geopolitical events directly impacted the motivations, capabilities, and intentions of cyber threat actors, including nation-state cyber capabilities, financially motivated DDW actors, ideologically motivated hacktivist collectives, and politically motivated activist groups. The dynamic and unpredictable geopolitical environment is expected to further influence cyber threat activities in 2025. Past and ongoing geopolitical events, such as the Russia-Ukraine war and the Israel-Hamas conflict, have facilitated elevated cyber threat activity. In 2025, we anticipate continued politically motivated cyber threats, including social engineering, data breaches, DDoS attacks, and malicious payload deployment, such as R&DE and spyware. Cybercriminal collectives are likely to align with geopolitical disputes, contributing to the complexity of the threat landscape. The EU's investment in high-tech fields and the geopolitical tensions between China, the US, and the EU are likely to intensify cyber threats, with state-backed actors targeting critical infrastructure and technology sectors. Russia and Iran are expected to use hybrid tactics, including cyber warfare, to advance their geopolitical agendas, further shaping the cyber threat landscape in 2025. — Adam Darrah, VP of intelligence, ZeroFox Initial Access Brokers (IABs) In 2025, Initial Access Brokers (IABs) are expected to remain a significant threat to organizations globally. The market for illicit network access surged in 2024, with record levels of IAB sales identified across DDW marketplaces. We anticipate this thriving market will continue in 2025, with IABs targeting organizations of all sizes, industries, and geographies. IABs sell unauthorized access to corporate networks by marketing compromised credentials and network entry points, allowing buyers to quickly exploit compromised networks with minimal investment and risk. The average purchase price of IAB sales in 2024 was under USD 5,000, offering substantial returns for threat actors, including R&DE collectives. The value of compromised access varies based on factors like information criticality, privilege level, and exploitation potential within the supply chain. Illicit access sales will likely continue underpinning the threat from R&DE operators in 2025, with security teams needing to be vigilant of IABs targeting them directly and indirectly via upstream partners. IABs are expected to focus more on third-party providers, perceiving them as having weaker security postures. North America will likely remain the primary target, followed by Europe, with industries like manufacturing, professional services, technology, retail, and financial services being the most attractive targets. — Adam Darrah, VP of intelligence, ZeroFox First Major AI-Generated Code Vulnerability Development teams have eagerly embraced AI, particularly GenAI, to accelerate coding and drive efficiency. While the push for the "10x developer" is transforming software creation, the need for speed can sideline or shortcut traditional practices like code reviews, raising significant security concerns. In the coming year, overconfidence in AI's capabilities could lead to vulnerable or malicious code slipping into production. GenAI is powerful but fallible — it can be tricked with prompts and is prone to hallucinations. This risk is not hypothetical: 78% of security leaders believe AI-generated code will lead to a major security reckoning. The CrowdStrike outage illustrated how quickly unvetted code can escalate into a crisis. With AI-generated code on the rise, organizations must authenticate all code, applications, and workloads by verifying their identity. Code signing will become an even greater cornerstone in 2025, ensuring code comes from trusted sources, remains unchanged, and is approved for use. Yet, challenges persist: 83% of security leaders report developers already use AI to generate code, and 57% say it's now common practice. Despite this, 72% feel pressured to allow AI to stay competitive, while 63% have considered banning it due to security risks. Balancing innovation with security will be critical moving forward. — Kevin Bocek, chief innovation officer, Venafi , a CyberArk company Everyone's a Creator: The Democratization of Specialized Knowledge Work In 2025, AI tools will revolutionize knowledge work by enabling individuals to tackle tasks once reserved for specialists, from coding to design and content creation. Much like personal computers empowered workers to handle spreadsheets and documents independently rather than relying on centralized admin staff, AI will push creativity and productivity to the edge, placing advanced capabilities in the hands of individual contributors. This shift will not only accelerate workflows but also challenge traditional organizational structures as more people leverage AI to go solo or create in new ways. AI's role as a personal assistant and creative partner will reshape industries, making innovation more accessible than ever before. — Rob Brazier, VP of Product, Apollo GraphQL AI-Driven APIs: A Wild Frontier In 2025, the relationship between AI and APIs will enter uncharted territory, reshaping how systems are built and interact. AI will increasingly guide developers in crafting and consuming APIs, introducing new patterns and unpredictable usage scenarios. This shift will demand advanced observability tools to monitor and adapt to evolving behaviors, ensuring systems remain secure and efficient. As AI dynamically composes user experiences in real time, APIs will need to be more robust, resilient, and flexible than ever before. Businesses must embrace this wild frontier with innovation and foresight, as the synergy between AI and APIs transforms digital ecosystems in ways we're only beginning to understand. — Rob Brazier, VP of Product, Apollo GraphQL AI and APIs: The Backbone of Intelligent Innovation In 2025, the fusion of AI and APIs will redefine how businesses build and run intelligent systems. APIs will evolve from simple connectors to dynamic engines for innovation, driving experimentation and production at unprecedented scales. As AI applications proliferate, organizations will demand APIs that not only handle the chaos of rapid prototyping but also balance speed with robust security and cost efficiency in production environments. Granular access controls, real-time performance monitoring, and optimized compute environments will become non-negotiable for businesses navigating this new era. APIs will act as the trusted gatekeepers of sensitive data, ensuring that AI-driven systems are both powerful, smart, and secure. This synergy between AI and APIs will empower developers to build smarter, faster, and more resilient applications, setting a new standard for innovation across industries. — Subrata Chakrabarti, VP of Product Marketing at Apollo GraphQL Smarter AI for Specialized Needs In 2025, the future of AI will shift toward smaller, domain-specific systems designed to excel in targeted applications. These compact, context-rich models will redefine industries by offering unparalleled efficiency and precision. Rather than relying on broad, generalized AI, businesses will start to adopt solutions tailored to their unique needs — healthcare organizations will use AI for diagnostics, while financial institutions enhance fraud detection. By embedding deep and specialized knowledge directly into models, companies will deliver real-time insights and reduce resource demands. AI will take a step forward towards decision-making, serving as a critical assistant rather than a complete solution. This evolution will make AI more practical, accessible, and impactful, transforming specialized knowledge from an advantage into a necessity. — Subrata Chakrabarti, VP of Product Marketing at Apollo GraphQL Ethics in AI Will Take a Step Forward in 2025 In 2025, geopolitical turbulence will continue, and misinformation is likely to abound. It's unlikely that new data privacy and AI policies will be passed and enforced in 2025, so customers will expect businesses to take responsibility for ethics in AI. As companies incorporate AI into their products, they have a responsibility to protect what and how the AI uses customer data, especially as it relates to sensitive data. Businesses must invest in ethical AI development, with an emphasis on transparency because AI adoption will directly correlate to the amount of trust the customers have in it. — Stephen Manley, CTO, Druva 2025 Will See the First Data Breach of an AI Model Pundits have frequently warned about the data risks in AI models. If the training data is compromised, entire systems can be exploited. While it is difficult to attack the large language models (LLMs) used in tools like ChatGPT, the rise of lower-cost, more targeted small language models (SLM) make them a target. The impact of a corrupt SLM in 2025 will be massive because consumers won't make a distinction between LLMs and SLMs. The breach will spur the development of new regulations and guardrails to protect customers. — Stephen Manley, CTO, Druva Synthetic Data Used More in AI Training to Safeguard Sensitive Customer Data, Creating New Risks For AI to produce good results, it needs to be trained on good data and rigorously tested with prompt engineering. The business temptation is to use customer data to train AI models — but that causes a myriad of problems to crop up, such as data compliance breaches, higher impact of cyber risk, and higher likelihood of data leakage. To effectively combat these challenges, businesses will turn to synthetic data, or training data that AI models generate, to maintain safety best practices during the training process. This, however, will create new risks, since the synthetic data can create a feedback loop that will exacerbate any bias in the data. Therefore, companies will need to invest in transparency and increase the rigor in reviewing their AI-generated output. — Stephen Manley, CTO, Druva 2025 Is the Year of (Missing) ROI on GenAI Investments The trough of disillusionment looms for GenAI, and the request for ROI will quicken the industry's descent into said trough. Every business is striving to understand the impact of GenAI, and savvy business leaders are already asking questions around accuracy, efficiency, and outcome to validate the IT spend allocated to it. Unless it's incorporated into a purpose-built tool from the ground up, GenAI won't drive significant measurable efficiency and many will feel let down by its initial promises. — Stephen Manley, CTO, Druva Security Leaders Will Embrace AI Experimentation 2024 shocked many of us with AI technologies' sophistication and rapid advancement. The year also highlighted that we don't quite know how to incorporate such tools into work and which vendors can help us along the way. Organizations in 2025 will continue to experiment with AI to understand where it offers value. And we'll also see many startups experiment with business models and tech approaches. Security and IT leaders should be ready to help evaluate and onboard a diverse set of immature AI products. We'll need to comprehend a range of AI technologies and understand the expectations of diverse internal stakeholders so we can contribute toward making informed risk vs. reward decisions. — Lenny Zeltser, SANS Institute Fellow and CISO at Axonius AI in Security: Balancing Human Expertise and Automation for Optimal Outcomes AI-related advancements will continue to fuel discussions regarding the role of humans vs. automation in the workforce. Security teams will see more opportunities to use AI and non-AI technologies to automate tasks across many domains, including GRC, security operations, and product security. Security leaders will need to be strategic about deciding which tasks to leave for humans and which to automate. Given how rapidly the technology is changing, we should be ready to experiment and determine how to measure project outcomes to decide which approaches work best. — Lenny Zeltser, SANS Institute Fellow and CISO at Axonius Multi-agent Neurosymbolic AI Will Advance Machine-to-Machine Collaboration The first wave of multi-agent neurosymbolic AI applications that perform machine-to-machine collaboration will emerge in 2025. Agents across diverse systems — such as autonomous vehicles, robotics, and enterprise decision support platforms — will exchange and interpret complex symbolic representations of their surroundings in real time. These agents will work together to negotiate solutions, adapt to new situations, and coordinate actions based on both learned experiences and structured knowledge. This advancement will lead to a new wave of AI products capable of more intelligent teamwork and enhanced performance in complex environments, all while ensuring transparency and explainability in decision-making. — Dr. Jans Aasman, CEO, Franz 2025 Will Be the Year of the AI Agent Instead of merely producing text or images, this new breed of AI application will be empowered to act. That might mean researching topics on the web, manipulating an application on a PC desktop, or any other task that can be performed via API. We're still a long way from general artificial intelligence, so these early agents will be quite specialized. We'll see the emergence of what might be called "agentic architectures" — focused use cases where AI can deliver immediate value. Likely examples include data modeling, master data management, analytics and data enrichment, where tasks are highly structured and prototypes have already shown promise. We'll see the first case studies in 2025, and then rapid uptake throughout the enterprise as lagging adopters see competitors gaining an edge. — Bob van Luijt, CEO, Weaviate AI Moves Closer to the Edge In the year ahead, we anticipate AI at the edge will further enhance applications and improve efficiency with increasingly specialized edge-AI chips that can enable tasks with lower power consumption. AI techniques like TinyML and model quantization will continue to advance, allowing more sophisticated AI algorithms to run on resource-constrained devices. We expect more real-time speech recognition, computer vision, and predictive maintenance on small edge devices, along with more local data processing. Current edge applications mostly use pre-trained models, but a move toward real-time, on-device training and fine-tuning will become more common. This means edge devices could adapt and learn from local data over time, improving performance and personalization without relying on cloud retraining. — Rashmi Misra, chief AI officer, Analog Devices Business Leaders Must Measure Value of AI Apps Companies that rush into AI adoption without understanding their internal needs and bandwidth risk overwhelming their security and data teams with information that doesn't provide valuable insights. As AI continues to grow, businesses aiming for long-term ROI must shift their focus from simply integrating AI capabilities to addressing organizations' shortcomings and measuring value. To accomplish this in the coming year, business leaders should collaborate closely with internal teams to identify their processes, bottlenecks, and needs. By understanding these challenges, leaders can work strategically with their teams to determine the most effective AI applications and ensure their teams are prepared to manage them successfully. — Rishi Kaushal , CIO, Entrust The 'AI Winter' Is Not Coming We're currently experiencing one of the most sustained stretches of interest and investment in AI that we've ever seen. While traditionally we've seen this hype give way to "AI winters" where enthusiasm and funding taper off, this time around, there are strong indicators that this momentum will continue into the new year and beyond. 2025 will be a year where scaled production of AI will sustain the investment in AI for years to come. This is just the beginning. — Raj Pai, Vice President, Product Management, Cloud AI, Google Cloud 2025 Is the Year of the Platform If 2024 was the year of the LLM, 2025 will be the year of the platform. There's no shortage of models on the market — plenty to address just about any use case. But there's no point for businesses in talking about models if you don't have a strong platform to support them. In 2025, technology leaders will shift their focus toward investing in platforms that have built-in security, grounding capabilities to reduce hallucinations, and can serve as a one-stop-shop to bring the potential of these models to life. — Raj Pai, Vice President, Product Management, Cloud AI, Google Cloud AI Model Convergence Will Continue If we look at the last 10 years of deep learning — now referred to as AI — we have been on a path of convergence. For example: In earlier days, we had separate models for different tasks like sentiment analysis, parts of speech tagging, and entity detection. But with models like BERT, a single model started performing all these tasks. Similarly: For translation, there were individual models for translating each language pair (i.e. English to Spanish, French to German, etc.). Now, a single model can translate across any pair of hundreds of languages. As we head into 2025, we'll see this convergence trend continue with things like screen understanding and reasoning, where a single model will have the power to do multiple tasks of varied nature, modalities, and across languages. With this strong technology march toward convergence, useful agentic behavior will natively start showing up across the different foundation models. — Saurabh Tiwary, VP, General Manager, Cloud AI, Google Cloud Strengthening Cybersecurity Against AI-Generated Threats With escalating threats from sophisticated phishing and ransomware attacks, focus needs to shift toward advanced data protection strategies, AI-driven threat detection and continuous employee training to mitigate ongoing risks. Businesses that proactively adopt these measures will not only comply with regulations but also build customer trust and loyalty. — James Tommey, Global Head of IT & Chief Security Officer, DISCO Combating Fraudulent AI-Generated Content In 2025, organizations will face unprecedented cybersecurity challenges due to the rise of fraudulent AI-generated content, which will become indistinguishable from human-created data. Leaders must think about how to implement robust authentication and verification protocols to safeguard against deepfakes and synthetic data breaches to ensure protection over the integrity of their workflows. — James Tommey, Global Head of IT & Chief Security Officer, DISCO AI PCs Will Be a Hot Commodity As the current PC refresh continues, AI PCs will be the primary choice in 2025 and beyond. We are only in the beginning stages of unlocking new workforce collaboration, security, productivity and even fulfillment through AI. For example, AI PCs now support real-time translation, extending global connection and collaboration seamlessly. As video communication cements itself as the norm, new eye-tracking capabilities allow participants to maintain perceived eye contact while focusing on facial cues and on-screen content, creating a more personal interaction. As business leaders evaluate their investments in new technology, AI PCs are a clear choice to begin equipping their workforce for the future with the high-performance computing required to scale and support increasingly data-driven and AI-powered work. — Dave McQuarrie, Chief Commercial Officer, HP Inc. Breaking Down Data Silos Will Become a Central Focus for AI and Data Architects In 2025, breaking down data silos will emerge as a critical architectural concern for data engineers and AI architects. The ability to aggregate and unify disparate data sets across organizations will be essential for driving advanced analytics, AI, and machine learning initiatives. As the volume and diversity of data sources continue to grow, overcoming these silos will be crucial for enabling the holistic insights and decision-making that modern AI systems demand. The focus will shift from the infrastructure toward seamless data integration across various platforms, teams, and geographies. The goal will be to create an ecosystem where data is easily accessible, shareable, and actionable across all domains. Expect to see new tools and frameworks aimed at simplifying data integration and fostering greater collaboration across traditionally siloed environments. — Molly Presley, SVP of Global Marketing, Hammerspace GPU Demand Soars, but AI Adoption Has Companies Rethink Resource Allocation As we enter 2025, the AI industry faces an unexpected situation: a huge demand for GPUs worldwide, yet many of these powerful chips aren't being fully used. While companies invested heavily in GPU-based infrastructure, many continue to struggle to apply these chips to AI workloads, instead redirecting them toward non-AI applications. The expected AI-driven boom remains slower than anticipated. We will continue to see companies be more selective with GPU allocations, as companies focus on areas where the impact of AI in areas like data analytics and cloud computing enhancements — rather than emerging AI initiatives. Additionally, as developers become more resource-conscious, the focus on optimizing algorithms for available hardware, leveraging CPU-bound AI, and adopting hybrid approaches could become central trends. Ultimately, 2025 may be a year that companies will adapt to both the technical and logistical challenges of realizing AI's potential. — Molly Presley, SVP of Global Marketing, Hammerspace Generative AI in 2025: A New Era of Innovation As we move into the new year, I'm excited to see generative AI continue its rapid evolution, especially in areas where progress is already accelerating. Models focused on code and math (anything with well-defined reward signals) will become even more capable, pushing the boundaries of what we can automate and optimize. I expect open-weight models to reach a level of performance that makes them viable for a wide range of practical applications, making cutting-edge AI more accessible than ever before. Another area to watch is the growing role of AI-generated audio and video content. We will soon see this kind of content becoming a significant part of our everyday media consumption. I believe we're on the cusp of a major scientific breakthrough driven by AI, which will have profound implications for research and innovation. The pace of progress in generative AI is only going to accelerate, and I can't wait to see where it takes us next. — Percy Liang, co-founder, Together AI Agentic AI Will Take Center Stage, Delivering on Personalization and Efficiency In 2025, AI won't just be a tool; it will be a collaborator. Many AI-powered tools in use today are based on static rules or datasets. Agentic AI differs in that it can continuously learn from user inputs and integrate contextual information (think: account history, network environment, user behavior patterns and preferences), and make decisions with little to no human oversight. In other words, unlike today's approaches that require user prompts or predefined rules, agentic AI will operate proactively. Imagine a customer service AI that predicts user needs before a query is made, or a network management AI that identifies potential issues and resolves them autonomously, ensuring uninterrupted service. These AI agents will not just interact with humans or devices directly, but will also be able to discover, learn, and collaborate with each other to form complex workflows and/or chains of operations to automate even advanced business functions. For instance, multiple AI agents could automate supply chain management by coordinating with each other to forecast demand, optimize inventories, coordinate deliveries, and even negotiate with suppliers. For businesses, this shift means a leap in efficiency and personalization. It also underscores the importance of governance and guardrails. In response to the rise of agentic AI, we will see organizations implementing mandatory ethical guidelines to ensure fairness and transparency in algorithmic decisions and protecting intellectual property. — Liz Centoni, Executive Vice President and Chief Customer Experience Officer, Cisco AI Will Surface Tough Reality Checks for Companies AI will continue to captivate businesses, promising unprecedented innovation and efficiency, and companies will continue to invest in AI-powered solutions. This is hardly a prediction. But as AI journeys progress, so too will the understanding that the path is fraught with hurdles. Despite billions of dollars invested into AI models and AI-powered solutions in 2024, new data from Cisco's AI Readiness Index shows that AI readiness has declined by one point globally over the past year — now only 13% of companies are ready to leverage AI-powered technologies to their full potential. In 2025 organizations will grapple with how best to secure the right level of compute power to meet AI workloads (today, only 21% of organizations say they have the necessary GPUs to meet current and future AI demands). Companies will need to lean on their strategic partners to identify and prioritize their AI use cases, upskill their teams, and modernize their infrastructure environments in a progressive, proportional way. IT teams will experience increasing pressure to optimize the management, hygiene, labelling, and organization of data, which is currently spread across multiple systems and locations. This mandate will apply to structured data typically associated with existing business processes, as well as unstructured data related to customer and user interactions. As teams work feverishly to prepare their environments for AI, boards and leadership teams will realize that significant gains from AI will happen in the long run and progressively — starting now and improving over time — especially in areas like opening new revenue streams and improving profitability. Many boards will find themselves readjusting expectations, timelines and priorities that were established mere months ago as companies reckon with the "messy middle" of AI implementation. Let's play the "long game." — Liz Centoni, Executive Vice President and Chief Customer Experience Officer, Cisco Companies Will Need Help to Balance Sustainability and Growth in an AI-Powered Era The environmental impact of AI is the elephant in a lot of rooms. AI requires high energy consumption levels that impact carbon emissions across the board. By 2025, the amount of energy used by data centers dedicated to AI is expected to match the amount consumed by a country the size of the Netherlands in one year. Indeed, in many of my AI conversations with customers, sustainability emerges as a core concern. In 2025, customers will increasingly seek out partners who can deploy technology while helping them meet their net-zero commitments and sustainability goals on their current timeline. Businesses that win will be those who prioritize energy-efficient products and circularity in business models. Interestingly, AI-powered technology could also play a crucial role in unlocking energy efficiencies. Businesses will see AI unlock a new era of "energy networking" that combines software-defined networking capabilities with an electric power system made up of direct current (DC) micro grids to deliver more visibility into emissions, and a platform for optimizing power usage, distribution, and storage. In 2025, AI will be both the "what" and the "how" in this space, bringing us vast capabilities and a continuous learning method for delivering them more sustainably. — Liz Centoni, Executive Vice President and Chief Customer Experience Officer, Cisco AI Inches Closer to the Edge 2025 will be the year of real-time, multimodal AI. AI will enter the action with humans and machinery in entirely new ways — from bringing data from sensors, drones, robotics and machinery all together to take action. — Dan Wright, CEO and co-founder, Armada Energy and Defense Reach an AI Tipping Point In 2025, AI will hit its tipping point in energy, with edge computing bringing intelligence directly to the oil rig. Much like how railroads revolutionized the oil industry by unlocking new markets in the 19th century, cutting-edge computing infrastructure will transport AI to the farthest reaches of the edge in the 22nd century. 2025 will also mark a seismic shift in defense, as edge computing becomes indispensable in the era of autonomous warfare. It's the modern-day railroad that delivers AI to the frontlines, empowering the U.S. military to navigate the complexities of the battlefield with unprecedented speed and precision. — Dan Wright, CEO and co-founder, Armada AI Will Enhance Customer Experience Management This is the area where most companies are beginning their GenAI journey. They are trying out this new technology in a low-risk area to start. By offloading repetitive tasks requiring simple answers or informational lookups, companies seek to boost the customer experience with faster, more detailed answers through GenAI and RAG. — Adrienne Wilson, Director of Sales, Esker Organizations Will Increasingly Use AI to Meet Sustainability Goals In certain locations, companies are required to report on their sustainability results. By leveraging AI, automated suggestions can be made to choose a more sustainable option when procuring goods. Collecting and utilizing this data will help companies to meet these requirements. — Adrienne Wilson, Director of Sales, Esker Business Leaders Develop More Mature AI Assessment Procedures To enable business leaders to more effectively cope with the onslaught of "AI enabled" tools — and to minimize an oversight bottleneck — the industry will need to develop a set of foundational rubrics to guide in more timely assessments of AI technologies. As a result, I predict we will see a renewed focus on data classification labels, a better understanding of AI processing locations, and a demand for confidentiality assertions from vendors as private data traverses their infrastructure. As the industry transitions to an application-driven phase of AI, it is imperative that organizations be equipped to make thoughtful and timely decisions about how the technology can be used responsibly to drive business objectives. — Michael Covington, Vice President of Portfolio Strategy, Jamf Schools Reinvigorate Efforts to Protect Students Online in the Wake of AI Proliferation We'll see a strong push for more safety mechanisms to be installed on student devices, specifically when it comes to data protection, threat prevention, and privacy controls. Educational institutions will be encouraged (or perhaps required) to improve encryption protocols and access controls, use AI-powered threat detection to fight AI-powered attacks, use systems that provide real-time alerts, and step up their game when it comes to student data privacy. — Suraj Mohandas, Vice President, Strategy, Jamf Enterprises Get a Reality Check on the Value of GenAI Household name companies in cybersecurity to small new startups with 10 employees have quickly entered the GenAI market over the past year or two. It's a crowded space that can easily overwhelm even leaders of technology companies who are looking to select the right GenAI solution for their businesses. In 2025, while the hype cycle for GenAI will continue to evolve, we'll see the more effective solutions surface and more customers focusing on solutions that bring the most real value to their businesses. As with any "hot new tech" on the block, the buzz around this latest emerging technology will start to calm, and we'll start to see GenAI mature. We'll start to see what value these tools can provide for businesses, and which perform better than the others. It's going to be a year of cutting through the GenAI noise, and those who can break through that will be the companies that stick around for years to come. — Linh Lam, CIO, Jamf To Open Source AI or Not? Navigating Innovation and Security Challenges Open-source AI opens the door to unparalleled collaboration and innovation, but it also forces us to grapple with security, transparency, and trustworthiness questions. Organizations must weigh the benefits of openness against the potential risks of exposure. — Balaji Ganesan, co-founder and CEO, Privacera Continued Proliferation of AI Use Cases While AI isn't new, the momentum behind it is unprecedented. In 2025, we'll see a proliferation of AI use cases that redefine business processes. It's a transformative moment — some view it as a job risk, while others embrace it as an opportunity to innovate and thrive. — Sascha Giese, Global Tech Evangelist, Observability, SolarWinds Balancing Innovation and Regulation: The Rise of Responsible AI Policies Regulatory frameworks are stepping in to define the ethical, secure, and responsible path forward for AI and data usage. This is a wake-up call for organizations — compliance must transform from a checkbox exercise to a differentiating value proposition. Embracing these standards involves legal alignment and leading with purpose and integrity. — Balaji Ganesan, co-founder and CEO, Privacera AI-Powered Predictive Maintenance and Risk Management to Dominate Building Systems Managed services that monitor and optimize physical assets throughout their lifecycle will be table stakes. This includes critical functions like firmware updates, system health monitoring, and ensuring proper functionality. Predictive maintenance powered by AI will play a pivotal role in addressing vulnerabilities proactively, minimizing downtime and costs while bolstering security. The growing interconnectivity of building management systems brings new risks, including unvetted device access and limited visibility into system components. In 2025, facility managers need a layered risk management strategy that incorporates tiered system criticality, comprehensive remediation plans, and continuous auditing. — Greg Parker, Global Vice President, Security and Fire, Life Cycle Management, Johnson Controls AI Will Bloom as Organizations Shift Focus to ROI and Efficiency We've already seen all of the stages of the beginning "new technology" cycle with AI. In 2023, organizations were exploring and experimenting, and in 2024, they were implementing AI at scale. Because of the widespread implementation, in 2025, we will see an emphasis on ROI, and a focus on how AI can enable more efficient work. At this point, organizations should have overcome the initial challenges with AI, and so 2025 will be the year of letting it loose and seeing it bloom. — Jen Chew, VP Solutions & Consulting, Bristlecone AI and Automation Will Take Over Tedious Vulnerability Management Tasks Security teams are overwhelmed by the growing volume and complexity of vulnerabilities, leading to errors and burnout. AI-driven tools are set to change this, automating tasks like triage, validation, and patching. By analyzing vast datasets, these tools will predict which vulnerabilities are most likely to be exploited, allowing teams to focus on critical threats. By 2025, up to 60% of these tasks will be automated, significantly improving accuracy and response times. AI-driven tools will also proactively discover vulnerabilities, closing gaps before attackers can exploit them. — Jimmy Mesta, CTO and founder, RAD Security AI Will Give CISOs and Security Teams a Head Start on Threats It's no longer enough to detect threats after they've infiltrated a system. By training models on vast amounts of historical data, AI will help security teams spot emerging attack patterns before they cause damage. By detecting subtle anomalies in network traffic and user behavior, AI will provide proactive alerts, giving organizations a critical edge. This approach could cut the average time to detect threats (MTTD) by half. Moreover, as AI continues to advance, multi-agent systems will emerge as a new challenge. Attackers will use these systems to orchestrate sophisticated, automated attacks, forcing defenders to adopt similarly sophisticated AI solutions. — Jimmy Mesta, CTO and founder, RAD Security AI Will Help Close the Cybersecurity Skills Gap The demand for cybersecurity talent keeps growing, but there aren't enough skilled professionals to fill the gap. AI-powered tools are stepping in to level the playing field, helping organizations of all sizes automate threat detection, incident response, and compliance tasks. In the new year, over half of small and medium-sized businesses will depend on AI to manage their security operations. These tools will make advanced protection accessible, especially for teams with limited resources. — Jimmy Mesta, CTO and founder, RAD Security AI-Driven Threat Detection Will Integrate Seamlessly into DevOps Workflows AI will become fully integrated into DevOps workflows, enabling security to be embedded directly into the development process. With cloud-native environments growing more complex, AI-powered threat detection will continuously monitor applications in real-time, catching vulnerabilities before they can escalate. Rather than interrupting development cycles, AI tools will seamlessly provide proactive alerts and insights, helping teams address security issues as they arise — without slowing down the pace of innovation or deployment. — Jimmy Mesta, CTO and founder, RAD Security AI Will Simplify Compliance in an Era of Stricter Regulations As global data privacy and cybersecurity regulations become stricter, compliance will become an even more significant challenge. Traditional, manual compliance processes won't be enough anymore. By 2025, AI will automate compliance workflows, including auditing, reporting, and monitoring regulatory requirements in real-time. AI tools will identify gaps, generate actionable insights, and help organizations stay agile in adapting to evolving legal landscapes, freeing up security teams to focus on proactive protection. — Jimmy Mesta, CTO and founder, RAD Security AI Workload Security Will Address New Attack Vectors As AI becomes central to operations, attackers are targeting foundational elements like training datasets, where a single compromise can create widespread vulnerabilities. AI workload security will be crucial, focusing on protecting models from data poisoning, model evasion, and adversarial attacks. By 2025, integrated security solutions will safeguard AI throughout its lifecycle, ensuring data integrity and resistance to tampering. — Jimmy Mesta, CTO and founder, RAD Security Agentic AI Will Transform into Agentic Workflows That Drive Exponential Efficiency and Innovation As the adoption of agentic AI ramps up, we will also continue to extend it. In 2025, the technology will become mature enough to have multiple AI agents work together and feed into each other to orchestrate multi-step objectives — they will transform into agentic workflows that tie into each other to make decisions and perform more complex enterprise tasks. With agentic workflows, your systems will retain memory and intelligence and will have a high degree of adaptability in order to proactively adjust workflows based on the environment's responses. This multiplies efficiency and innovation exponentially. — Abhinav Puri, VP of Portfolio Solutions & Services, SUSE For Executives, Optimizing AI Cost and Performance Will Require a Strategic Balancing Act Firstly, identifying high-impact use cases will be crucial. This means prioritizing AI initiatives that directly contribute to core business objectives and offer measurable ROI, such as automating critical processes, enhancing customer experiences, or optimizing supply chains. Investing in robust data infrastructure and efficient AI models will also be key, ensuring the foundation for accurate and reliable AI-powered solutions. Secondly, embracing efficient AI practices will be essential. This includes leveraging solutions for scalability and cost-effectiveness, ensuring effective GPU utilization, fine-tuning AI models to reduce computational demands, and exploring techniques like model compression and knowledge distillation to optimize performance without sacrificing accuracy. By adopting a data-driven approach and continuously monitoring AI initiatives, executives can ensure they maximize the value of their AI investments while controlling costs. — Abhinav Puri, VP of Portfolio Solutions & Services, SUSE Multi-modal AI Is Set to Revolutionize AI in 2025 Multi-modal AI will enable machines to process and integrate information from multiple sources like text, images, video and audio. This breakthrough will lead to more intuitive human-computer interaction, enabling us to communicate with AI seamlessly using voice, gestures, and visuals. Imagine AI assistants that understand complex requests involving multiple forms of media, or robots that can perceive and navigate their environment with human-like awareness. Furthermore, multi-modal AI will fuel a wave of innovation across industries. Expect personalized learning experiences that adapt to individual needs, AI-powered tools that revolutionize content creation by generating videos from text or music from images, and advancements in healthcare with AI analyzing diverse patient data for accurate diagnoses. However, this progress necessitates a focus on ethical considerations, ensuring fairness and responsible use of these capabilities. — Abhinav Puri, VP of Portfolio Solutions & Services, SUSE Specialized Foundation Models Take Center Stage Implementation complexity: full coffee IV drip needed; market readiness: initial adoption; investment required: significant investment. While large language models (LLM) have dominated the conversation, the real innovation is happening in specialized foundation models. Look at what's happening in drug discovery, material science, and agriculture. We've already seen models predict over 200 million protein structures and discover 2.2 million new materials . In 2025, this universe of models will accelerate. Every major player has their own language model — that's becoming table stakes. The true differentiators will be these domain-specific models tackling complex scientific and mathematical challenges. — Vijoy Pandey, SVP of Cisco's incubation and innovation engine Outshift Show Me the Money: Agentic Apps Generate Revenue CIO sleep loss: Regular midnight thoughts. Developer excitement: Clear whiteboard needed. VC funding frenzy: Term sheets flying. Right now, everyone's experimenting with AI agents, but nobody's making real money yet. That changes in 2025. The foundational technology is ready — but we still need to solve core challenges around data quality, operational costs, and building trust. We need better ways for agents to communicate and collaborate. Think about something as simple as agentic creation of market analysis for a product - sounds straightforward, but nobody has deployed it yet. The market is ready for practical solutions that can demonstrate clear business value, ROI, and trust. — Vijoy Pandey, SVP of Cisco's incubation and innovation engine Outshift Agent Heterogeneity and the Sprawl Challenge Complexity: Counting grains of sand. Stack Overflow questions: "Please help!" flood. Enterprise FOMO: "Quick, schedule a meeting." We're heading into a world of "agent heterogeneity" — different vendors, different capabilities, minimal standardization which will create a growing challenge: agent sprawl. As AI gets built into every application and service, organizations will find themselves managing hundreds or thousands of discrete agents. Without open standards and frameworks, this diversity creates chaos. It's like the early days of networking — we need common protocols and standards so these agents can discover, communicate, and collaborate with each other effectively. This standardization and interoperability will be essential for enterprises to effectively manage and scale their AI initiatives. — Vijoy Pandey, SVP of Cisco's incubation and innovation engine Outshift From Solo Tasks to End-to-End Processes Market readiness: Early experimentation. Industry disruption: Cross-industry transformation. Developer excitement: Keyboard literally smoking. Today's AI assistants are like solo performers — good at individual tasks like drafting emails or analyzing data. In 2025, we'll see the full orchestra — AI systems managing complex end-to-end business processes. Supply chains will be orchestrated by collaborating AI systems handling everything from demand forecasting to logistics optimization, all adapting in real time. The key is moving from isolated tasks to integrated workflows that deliver real business outcomes. — Vijoy Pandey, SVP of Cisco's incubation and innovation engine Outshift AI and ML to Revolutionize Retail Supply Chains As we move into 2025, AI and machine learning (ML) will reshape retail supply chains, driving efficiency and adaptability. More importantly, as the pace of product life cycles quickens, predictive analytics can help retailers anticipate shifts and restock faster, avoiding costly shortages or oversupply. From demand forecasting to personalized shopping experiences, technology is transforming retail at every touchpoint, enabling brands to build deeper connections and respond dynamically to consumer needs. — Keith Nealon, CEO, Bazaarvoice AI-Powered Personalization AI and machine learning are revolutionizing how brands engage with consumers. From personalized recommendations to automated customer service, these technologies offer insights and experiences at a scale that was previously impossible. And these experiences are what shoppers crave — according to our research, personalized offers drive 45% of shoppers to complete purchases online. In 2025, the brands that leverage AI to deliver hyper-personalized experiences and maintain a responsive, flexible supply chain will have a significant edge in building long-term customer loyalty. — Colin Bodell, Chief Technology Officer, Bazaarvoice The Rise of Autonomous Agents In 2024, we told AI what to do. In 2025, AI agents will start doing the actual work while we watch. Software engineers will input what they want, then see their screen come alive — the cursor moving on its own, opening files, writing code, running tests, fixing bugs. It's not about AI suggesting code in a chat box anymore. The cursor will literally move by itself, doing real development work. Microsoft and GitHub are already testing early versions. The change will be striking — from AI as a smart assistant to AI as a capable executor, handling full development workflows while engineers focus on higher-level decisions. — Andrew Feldman, Founder & CEO, Cerebras The Emergence of 'Thoughtful' AI Current AI is really just pattern matching — fast but shallow. 2025 brings something fundamentally different, sparked by OpenAI's O1 breakthrough in test-time computation. AI will start taking variable time to think through problems. Simple questions get instant answers. Complex system design questions? The AI will tell you "need a few minutes to think this through properly." It's a natural evolution — harder problems need more processing time. The implications are significant — AI that can tackle genuinely complex analytical work, taking more compute time when needed to generate better answers. — Andrew Feldman, Founder & CEO, Cerebras The End of Nvidia's AI Chip Monopoly The AI chip market is finally opening up beyond Nvidia's dominance. Cerebras is making real progress in high-performance inference, especially for companies running large language models. AWS and Google are rolling out chips optimized for cost-efficient AI workloads, while Apple is pushing AI computation to mobile devices. Nvidia will still sell every chip they make, but companies now have choices for different AI needs. The result? A more mature hardware ecosystem where different workloads can find their optimal chips — from edge devices to data centers. — Andrew Feldman, Founder & CEO, Cerebras AI Crossed 3 Billion Active Users AI is hitting growth curves we haven't seen since early mobile. We're at roughly 1 billion active users now — Meta's AI features reach 500M monthly actives, OpenAI has 300M weekly users, and that's not counting Google, Apple, and others. 2025 will see us cross 3 billion as AI becomes woven into the fabric of every major platform. The acceleration is driven by three forces: mobile AI getting serious with Apple and Android pushing on-device models, messaging apps making AI the default experience, and workplace tools embedding AI into daily workflows (think Microsoft Office, Google Workspace). When the core tools billions already use become AI-first, crossing 3B users isn't just possible — it's inevitable. — Andrew Feldman, Founder & CEO, Cerebras AI/ML Won't Replace Engineers but Give Them Superpowers While AI/ML continues to be an important emerging tool, we've moved beyond treating it as a catch-all solution. We've begun to hone in on specific use cases that deliver tangible value. AI/ML excels at pattern recognition, enabling the automation of time-consuming tasks and identifying cost-saving opportunities. Rather than replacing engineers, our solutions will augment their capabilities by reducing noise and providing well-reasoned recommendations. This frees up human experts to focus on complex decision-making where their expertise is most valuable. — Quynton Johnson, product marketing lead, Grafana Labs Digital Memory Curation In 2024, people were building reliance on generative AI tools like ChatGPT or Claude to answer questions and save time. In 2025, this will go a step further: AI will capture conversations and create insights to make you more productive. By finding patterns between conversations within meetings, calls, and videos, AI will start to establish a “digital memory” for its users. — Jason Chicola, CEO, Rev AI Job Opportunities AI will create a ton of new jobs just like the internet did years ago — some we can't even imagine yet. Positions like Prompt Engineer will start cropping up in 2025 as businesses focus more on AI ROI and push to see results. — Fernando Trueba, chief marketing officer, Rev Industry-Specific LLMs and Consolidated Productivity Suites In 2025, we'll see a shift toward industry-specific LLMs that are trained for specific sectors, and curated ecosystems of tools that integrate seamlessly across enterprises. These consolidated productivity suites will securely retain sensitive data and eliminate the need to repeatedly provide context, transforming AI from a novelty to essential business support. — Aron England, chief product and technology officer, Rev A Bigger Focus on AI ROI Companies spent an enormous amount of money on generative AI in 2024, but in many cases are still waiting to see its impact on top-line revenue. Although AI adoption has increased tremendously, we're still in the early stages where employees haven't quite mastered day-to-day use. In 2025, companies will draw a harsh line and say that if the AI tool isn't clearly contributing to ROI, it's gone. — Aron England, chief product and technology officer, Rev The Rise of AI Multimodals We'll see more quality maximalists — high-end models that can only run on a large server environment but can do it all. Single-purpose AI won't be as useful or prevalent as the dominant "omni" models start to emerge. For example, in the speech technology space, one model will take care of audio transcription, diarization, speaker labeling, and other needs in one fell swoop. — Lee Harris, VP of engineering, Rev Powerful On-Device AI No more sending your data to the cloud for outputs — more high-quality AI models will be available directly on mobile devices (phones, headsets, cars, etc. ), and usable without Internet service or a huge server in the background. This will be a boon for legal, law enforcement, journalism, and other professions where real-time availability and data security are key. — Lee Harris, VP of engineering, Rev AI Agent Cody Banks (just kidding) 2025 is set to be the year of the AI agent revolution. With exponential advancements in AI models and agentic workflows, AI agents will transform industries by automating complex tasks and enhancing efficiency. Major tech investments are accelerating this shift, making AI agents indispensable across business and daily life. The convergence of powerful technology and user-centric design will redefine productivity and innovation on a global scale. — Mike Diolosa, CTO, Qloo Consumers Will Expect Greater Personalization as Apple Intelligence Raises Expectations In 2024 we saw the rise of AI agents to assist humans in their day-to-day and enhance our efficiency. As AI agents become even more common, we'll start to see a rise in consumer demand for hyper-personalized experiences. At the same time, privacy-compliant consumer data will be essential for keeping on-device AI models truthful and valuable to end users. — Coby Santos, chief product officer, Qloo Production Studios and Content Companies Will Outgrow Their AI Fears Although 2024 saw controversy over the use of AI training data for content production, 2025 will see a boom in AI partnerships among big brands (such as Runway's recent announcement with Lionsgate ). Companies are starting to see the benefits AI can provide alongside protective AI guardrails to safeguard their existing human talent and customer data, and as a result they'll start fighting for exclusivity with the biggest tech brands to enhance their content and efficiency. — Alex Elias, CEO, Qloo Leaving Opt-out Defaults in 2024 Companies like LinkedIn and X have come under scrutiny for training AI models on consumer data by default, sometimes with no notification at all. Companies will revert to an opt-in strategy in 2025 and will ultimately protect their reputations by doing so, especially in a climate where consumer trust of AI is still not completely widespread. — Alex Elias, CEO, Qloo More Personalized AI-Powered Travel Recommendations AI-powered travel recommendations will grow more sophisticated and personalized, especially now that travelers prefer experiences that genuinely reflect their individual tastes and interests, instead of simply visiting the most popular destinations. On top of that, popular destinations are becoming undesirable as rising overtourism has created frustrated travelers and locals alike. — Jim Jansen, CRO, Qloo Real Estate Developers to Turn to AI AI is going to completely redefine what's considered prime real estate. Business districts in major cities are dying as empty office spaces and a slow worker return have culminated in a 52% drop in office value. In 2025, real estate developers will turn to AI to provide hyper-specific recommendations for restaurants and shops in these areas, turning these declining neighborhoods back into popular areas for consumers based on their unique taste profiles. — Levi Nitzberg, SVP of growth, Qloo GenAI Hype Cycle Comes Back Down to Earth Generative AI will never not be cool, but we reach a point where we give a slight nod to the hype cycle — and then get down to the business of delivering real value. This happens by simplifying our approaches, rules and models, complementing them with a targeted use of LLMs. Keep a close eye on that Nvidia stock. — Jared Peterson, Sr. VP, Platform Engineering, SAS Impacts of AI Regulation Regulation keeps AI in check but makes it challenging for businesses to use pure open source. Innovation takes a hit. Innovation silos crop up. AI enthusiasts cross their fingers that the impacts are temporary and hunker down to find solutions that work for their region. — Jared Peterson, Sr. VP, Platform Engineering, SAS The Titans of Tomorrow Are AI Augmented Today Fully AI-enabled organizations are the ones that will win the IT battles of 2025. As generative AI evolves from a "shiny new toy" to "just" another type of AI, organizations will fully operationalize AI to automate routine tasks that free employees for higher-value work. Those automations mean they'll make decisions faster, recognize opportunities more quickly, and drive more innovation than their competitors. In short: they'll win. — Jay Upchurch, chief information officer, SAS LLMs Get Commoditized … and Specialized In 2025, LLMs will become commoditized, leading to AI pricing models collapsing as base-level capabilities are offered for free. The real value will shift to specialized services and domain-specific applications built on top of these models. Simultaneously, the rise of open-source LLMs will challenge the dominance of a few key providers, driving a more decentralized AI landscape where customization and integration will be the key differentiators. — Udo Sglavo, VP, Applied AI & Modeling, R&D, SAS The Future Won't Be Right for Organizations That Fail to Act on Generative AI Think back to the digital transformation wave of the early 2000s. Companies that embraced the internet, digitized their processes, and invested in e-commerce became the Amazons, Googles, and Apples of today. Those organizations that waited or followed the wrong adoption path either adapted too late or disappeared entirely. Similarly, organizations that fail to act now will find it increasingly difficult to compete in the GenAI-powered economy. GenAI is not just another trend. It's the next leap in business evolution, and the organizations that understand this and move decisively will be the ones shaping the future. — Marinela Profi, global GenAI/AI strategy lead, SAS The Semantic Layer Becomes the Enabler for LLMs in Enterprises In 2025, the Semantic Layer will become the crucial enabler for LLMs in enterprises, acting as a bridge between internal data and LLMs to deliver precise, contextually relevant insights. By unifying enterprise data with global knowledge, this integration will revolutionize decision-making and productivity, making GenAI indispensable. Companies that embrace this convergence will dominate in innovation and customer experience, leaving competitors behind. — Ariel Katz, CEO, Sisense Ethical and Secure AI Takes Center Stage Companies will prioritize secure, customizable AI solutions that protect sensitive customer data while still leveraging the power of advanced analytics. AI governance frameworks will become essential for enterprises to ensure ethical use of AI in customer interactions and decision-making processes. Regulatory compliance in AI will drive innovation in transparent, explainable AI models for customer service applications. — Ashish Nagar, CEO of Level AI LLMs Will Hallucinate Much Less LLMs are still known to produce factually inaccurate or utterly random content, especially in languages other than English, where training data is sparse. There is a known phenomenon where the lexical coverage is sufficient to produce content in a foreign language. However, the actual cultural and social realia still stem from the English-speaking world or are entirely random. As the models are deterministic, they will produce an answer no matter what, even when the confidence level is low and despite prompts to respond with "I don't know." To combat this behavior, new techniques of detecting and mitigating hallucinations will continue evolving, opening the doors for more use cases where output reliability is vital. Such methods include analyzing correlations between edit rate, log probability, and semantic entropy, thus catching hallucinations and either following this analysis with a self-healing step, or sending potentially flawed content to a human-in-the-loop review; without a doubt, new approaches will be introduced both to prevent model hallucination and mitigate hallucinations in the post-processing step. — Olga Beregovaya, VP of AI, Smartling Stand-Out Innovative AI Applications The AI application that excites me the most is the idea of agentic AI — systems that can plan and execute tasks to meet goals that I define, assisting me in project planning, content creation, and ultimately making my life easier and or letting me focus on more creative tasks. — Jerod Johnson, Sr. technical evangelist, CData Organizations Shift to Targeted AI Initiatives Most organizations are moving towards AI-readiness. Recently I've noticed a narrowing of scope for AI projects — from an "AI will do everything" attitude to an "AI can help us/our customers in this very specific way." As orgs narrow their focus, they can create realistic goals for their AI initiatives and work backwards to determine how to build the models/training/etc. for their AI. A skill/infrastructure gap I see is effective access to data for every stakeholder in the organization. IT and developers have an easier time getting to data, because of their skillsets, but line of business users shouldn't be expected to know how to access data, while still being provided democratized access. — Jerod Johnson, Sr. technical evangelist, CData More Content Will Shift to 'No Human in the Loop' The quality of AI-generated content is getting exponentially higher, in many instances reaching human parity, especially for "not noisy," structured, professionally authored content types, like help systems, manuals, websites and eLearning content. About two or three years ago, the recommendation for such "branded" content would have been to use a "human in the loop" process. The global content transformation process is shifting towards instant, fully automated delivery. We will see more and more content types move towards "multilingual generation" (rather than traditional translation) using LLMs with RAG or few-shot examples and advanced prompt engineering. However, the adoption of this workflow will vary based on languages, simply due to the availability of model training data. Under-resourced languages will still require significant human effort for the generated or translated content to be at the level of quality needed. — Olga Beregovaya, VP of AI, Smartling Governance, Legal Frameworks, and Ethical Considerations Around AI Will Be More Structured and Transparent We could label 2023 as a year of "Generative AI Chaos," where there were more questions than answers when implementing AI-based technologies. The Infosec questionnaires and corporate or government guidelines were rather vague, and there was a lot of uncertainty about IP, data protection, PII handling and overall risk assessment. 2024 became a year of "measured deployment," where the learnings of AI implementations were being translated into standards and regulations. There are two aspects to such regulations: the ethics of actual deployments, where potential impact is analyzed and risks are mitigated, and the ethics of ensuring safety and emotional well-being of the workforce. We already see more local governmental regulatory bodies' initiatives around safe AI and such initiatives will have more global alignment in the fu