 |
|
|
|||||
Over the course of 2025, deepfakes improved dramatically. AI-generated faces, voices, and full-body performances that mimic real people increased in quality far beyond what even many experts expected would be the case just a few years ago. They were also increasingly used to deceive people. For many everyday scenariosespecially low-resolution video calls and media shared on social media platformstheir realism is now high enough to reliably fool nonexpert viewers. In practical terms, synthetic media have become indistinguishable from authentic recordings for ordinary people and, in some cases, even for institutions. And this surge is not limited to quality. The volume of deepfakes has grown explosively: Cybersecurity firm DeepStrike estimates an increase from roughly 500,000 online deepfakes in 2023 to about 8 million in 2025, with annual growth nearing 900%. Im a computer scientist who researches deepfakes and other synthetic media. From my vantage point, I see that the situation is likely to get worse in 2026 as deepfakes become synthetic performers capable of reacting to people in real time. Just about anyone can now make a deepfake video. Dramatic improvements Several technical shifts underlie this dramatic escalation. First, video realism made a significant leap, thanks to video generation models designed specifically to maintain temporal consistency. These models produce videos that have coherent motion, consistent identities of the people portrayed, and content that makes sense from one frame to the next. The models disentangle the information related to representing a persons identity from the information about motion so that the same motion can be mapped to different identities, or the same identity can have multiple types of motions. These models produce stable, coherent faces without the flicker, warping, or structural distortions around the eyes and jawline that once served as reliable forensic evidence of deepfakes. Second, voice cloning has crossed what I would call the indistinguishable threshold. A few seconds of audio now suffice to generate a convincing clonecomplete with natural intonation, rhythm, emphasis, emotion, pauses, and breathing noise. This capability is already fueling large-scale fraud. Some major retailers report receiving over 1,000 AI-generated scam calls per day. The perceptual tells that once gave away synthetic voices have largely disappeared. Third, consumer tools have pushed the technical barrier almost to zero. Upgrades from OpenAIs Sora 2 and Googles Veo 3 and a wave of startups mean that anyone can describe an idea, let a large language model such as OpenAIs ChatGPT or Googles Gemini draft a script, and generate polished audio-visual media in minutes. AI agents can automate the entire process. The capacity to generate coherent, storyline-driven deepfakes at a large scale has effectively been democratized. This combination of surging quantity and personas that are nearly indistinguishable from real humans creates serious challenges for detecting deepfakes, especially in a media environment where peoples attention is fragmented and content moves faster than it can be verified. There has already been real-world harmfrom misinformation to targeted harassment and financial scamsenabled by deepfakes that spread before people have a chance to realize whats happening. AI researcher Hany Farid explains how deepfakes work and how good theyre getting. The future is real time Looking forward, the trajectory for next year is clear: Deepfakes are moving toward real-time synthesis that can produce videos that closely resemble the nuances of a humans appearance, making it easier for them to evade detection systems. The frontier is shifting from static visual realism to temporal and behavioral coherence: models that generate live or near-live content rather than pre-rendered clips. Identity modeling is converging into unified systems that capture not just how a person looks, but how they move, sound, and speak across contexts. The result goes beyond this resembles person X, to this behaves like person X over time. I expect entire video-call participants to be synthesized in real time; interactive AI-driven actors whose faces, voices, and mannerisms adapt instantly to a prompt; and scammers deploying responsive avatars rather than fixed videos. As these capabilities mature, the perceptual gap between synthetic and authentic human media will continue to narrow. The meaningful line of defense will shift away from human judgment. Instead, it will depend on infrastructure-level protections. These include secure provenance, such as media signed cryptographically, and AI content tools that use the Coalition for Content Provenance and Authenticity specifications. It will also depend on multimodal forensic tools such as my labs Deepfake-o-Meter. Simply looking harder at pixels will no longer be adequate. Siwei Lyu is a professor of computer science and engineering and diector of the UB Media Forensic Lab at the University at Buffalo. This article is republished from The Conversation under a Creative Commons license. Read the original article.
Category:
E-Commerce
Strategic planning is a big business. Companies spend millions of dollars working with consulting firms to chart a path forward. Not only does a lot of money change hands as part of this process, but the amount of time that employees invest in working on the plan likely doubles the cost of the entire process. In the end, leadership gets a shiny report they can send to employees, shareholders, external stakeholders, and others. Often, though, much less money and time is invested in implementing that plan than was spent creating it. As a result, there is a lot of cynicism around engaging in strategic plans. In many ways, this feels a lot like New Year’s resolutions. With great fervor, people will identify a change they want to make in the new year. Now is the time to get physically fit, develop deeper relationships, or get an education. Yet, most people have abandoned their resolutions in a few weeks. The central problem with strategic plans is in the name itself. Every organization needs to be concerned both with strategy and tactics. Strategy defines the north star for the organization. What are the big-picture elements youre trying to accomplish? Tactics is the method for getting there. What specific steps are team members going to take on a daily, weekly, and monthly basis that will lead to the desired outcome. Ultimately, a strategy is unlikely to meet with success without a tactical plan to get there. There are several things leaders can do to increase the chances of success for a strategic plan. In many ways, these mirror the steps people need to take to be better at achieving their New Years resolutions. Focus on resources A big part of the problem with the strategic planning process is that the focus is almost entirely on strategy instead of the resources needed to execute on it. Organizations take their plan and then develop other teams tasked with turning that plan into a reality. This creates two central problems. There are inevitable tradeoffs that must be made to start to implement a plan, which dampens enthusiasm for the golden future the strategy promised. In addition, the resource (human, financial, and material) needed to implement the plan is rarely identified ahead of time, which leads to significant battles during implementation. A planning process should put most of the effort into the tactical planning rather than the strategic planning. Responsibility for particular elements of the plan should be given to specific groups. Money needed to move the plan forward should be identified early. The new work to be done should not just be dumped on top of the existing load carried by employees. Instead, responsibilities must be shifted so that people in the organization have the time to make progress on the new work. Otherwise, the plan will fail. Identify concrete steps If an organization is going to do things differently in the future than it does in the present, people are going to have to engage in different actions than they were before. That means you need to know what people are doing now. How do the actions people take now move the organizations mission forward? How can the elements of that mission that cannot be lost be integrated with tasks that will promote the new direction? Much of the success of this planning process also requires thinking through the reward structure for employees. In any organization, there is what you say, what you do, and what you reward, and people listen to those in reverse order. What you reward is what drives a lot of daily behavior. So, if you want people to do something different tomorrow than they were doing today, youre going to have to shift what people are rewarded for doing so that more of the actions related to the new goals is incorporated into the work day. This kind of specific exploration of the work day is not nearly as much fun as envisioning a bright future, which is why strategic planning processes often kick that can down the road. But, this kind of detailed work is directly related to the likelihood of success of the plan. Try, then adapt As Mike Tyson said, Everybody has a plan until they get punched in the mouth. The other reason that the planning process is fun (albeit unproductive) is that it is blissfully unsullied by reality. It is impossible to envision the issues that will inevitably arise as you implement a plan. Success at reaching a strategic goal is done in successive approximations. You try something, measure the outcomes, and then assess what is working and what is not. Keep what works, and fix what doesnt. Ultimately, your plans are more like software than hardware. Hardware is as good as it will ever be when it comes out of the box. Software gets better by patching the bugs and adding new features. When you commit to continuous improvement of your tactical plans, you greatly improve the likelihood of reaching strategic aims.
Category:
E-Commerce
Starting today, Google is weaving its massive investment in AI into one product nearly everyone already usesand for many people, the change wont feel optional. Google announced Thursday that a suite of new features powered by Gemini 3 will begin appearing in Gmail, introducing automation designed to reduce inbox overload. The most consequential update is a new Gmail view called AI Inbox, which reshapes email around summaries, topics, and to-dos, rather than individual messages. What changes the moment this turns on For users, the shift isnt about learning new toolsits about no longer having to manage email the same way. Instead of opening Gmail to a chronological list of messages, AI Inbox presents a briefing-style overview that surfaces conversations, tasks, and updates it thinks matter most. With email volume at an all-time high, managing your inbox and the flow of information has become as important as the emails themselves, Gmail VP of product Blake Barnes wrote in a blog post announcing the changes. Googles goal, he added, is to turn Gmail into a personal, proactive inbox assistant. The new AI Inbox wont roll out right away. Google says its currently testing the feature with a small subset of users, with a broader rollout planned for the coming months. Less searching, more trusting In general, the addition of AI is meant to make finding things easier. Google says Gmails new AI Inbox will offer a personalized briefing that prioritizes conversations based on how you use email, filtering out what it considers clutter so you can focus on whats important. In practice, that means relying less on Gmails search barand more on AI judgments about relevance. Thats a notable shift for a product used by roughly 3 billion people worldwide. Next to search, Gmail is Googles most ubiquitous service, functioning as the default archive for receipts, contracts, travel plans, conversations, and work history. Yet even as inboxes have grown more crowded, Gmails core experience has changed little. Google acknowledged that gap directly. Your inbox is full of important information, but accessing it has required you to become a power searcher, Barnes wrote. And even when you find the right emails, you are often left staring at a list of messages, forced to dig through the text to piece together the answer. The new approach aims to remove that burden entirely by summarizing, prioritizing, and contextualizing information before users ask for it. Your inbox as memory, not messages Every online interaction youve ever had likely lives somewhere in your inbox, but finding the right detail at the right moment has long required manual effort. With AI Inbox, Google wants to change that by treating Gmail less like a communication tool and more like an external memory systemone that can recall information, surface context, and suggest next steps. That idea aligns with how people increasingly use AI tools like ChatGPT, but applying the same concept to email raises higher stakes. Gmail doesnt just hold drafts and threads; it holds personal history. How well users trust AI-generated summariesand whether they stop opening original messages altogethermay ultimately determine whether the new interface sticks. Trust, not accuracy, is the real test The real test for Gmails AI makeover wont be whether its summaries are technically accurate, but whether users come to trust them enough to stop opening original messages at all. As AI-generated overviews begin to replace scrolling and searching, the act of verifying informationbe it reading an entire thread, checking dates, or scanning for nuancemay quietly fade. Over time, Gmail could train users to rely on interpretation rather than inspection, shifting email from a record people consult to a system they simply accept. Which features everyone getsand which they wont Many of the new AI-powered Gmail features will roll out to all users, but some of the most powerful tools will be reserved for paying subscribers. One widely available update, called AI Overview, summarizes long email threads and highlights key points, reducing the need to reread entire conversations. That feature is rolling out broadly. However, a more advanced capabilityasking Gmail questions like Who was the plumber that gave me a quote last year? and receiving an AI-generated answerwill only be available to subscribers on Google One AI Pro or Ultra plans, priced at $20 and $250 per month, respectively. For free users, Gmail becomes more readable. For paid users, it begins acting more like a searchable personal archive. Writing emails with less effort Google is also expanding AI tools designed to reduce the friction of replying and composing emails. A tool called Help Me Write, previously just an option for paid subscribers, will now be available to all Gmail users, along with “Suggested Replies,” a refresh of a tool previously called “Smart Replies.” Help Me Write will help users draft emails from scratch using prompts, while Suggested Replies generates a tailored one-click response based on the context of your conversation. Paid subscribers will also get access to “Proofread,” which offers more advanced grammar, tone, and style suggestions while composing messages. What youll need to opt out of Many of these features will be enabled by default, meaning users who prefer a more traditional Gmail experience will need to actively disable Gemini-powered tools in Gmails Smart Features settings. For those eager to hand off more inbox management to AI, the transition may feel overdue. For others, it may feel like Gmail has quietly crossed a linefrom organizing information to deciding what matters. Either way, once Gmail stops asking you to search your inbox and starts telling you what you need to know, email may never feel quite the same.
Category:
E-Commerce
All news |
||||||||||||||||||
|
||||||||||||||||||