Bots Now Outnumber Humans Online. Here’s What It Means for Your Business

It was only a matter of time before bots outnumbered humans on the internet, but many experts thought the flesh and blood majority would stand for a few more years. They were wrong—and the impact on business owners could be significant. New data from Cloudflare shows the number of bots accessing websites over the past seven days outnumbers human web users, with about 57 percent of web traffic coming from bots, who are busy browsing, querying, summarizing, shopping, researching, and scraping, increasingly via AI agents. “Welp, that happened faster than I predicted,” wrote Cloudflare CEO Matthew Prince in a social media post. “Thought it would be end of 2027, then early 2027, but agentic traffic [is] growing so fast that bots have now passed human traffic online for the first time in the internet’s history.” For business owners, that could mark the beginning of a new phase in how to handle business online. Instead of using the internet to attract human customers, it could be time to consider whether to structure your site to attract bots, say some experts. “Stop building for the human eye and start building for the machine mind,” says Rajiv Garg, a professor at Emory University’s Goizueta School of Business. “If an AI agent can’t read your data, you don’t exist.” The hurdle with that approach, though, is just as human visitors might either be customers or hackers looking for a weakness, not all bot traffic is the same. Some bots are malicious. Some are crawling for search engines. And some are sent from AI agents on behalf of potential customers. The challenge for businesses — and their tech teams — is figuring out which bots are which. “There is no protocol to verify whether an AI agent is acting on behalf of a real person, whether it has been authorized to perform its actions, or whether it is benign or hostile,” says Zach Meltzer, CEO and founder of Miami-based VeryAI, a ‘proof of reality’ platform designed to verify human identity and prevent AI-driven fraud. “Platforms cannot differentiate between a personal AI assistant booking a flight for its owner and a bot farm scraping data. The current workaround — forcing agents to impersonate humans via browser automation — is inefficient for legitimate agents and trivially bypassed by malicious ones.” There are also cost issues with bot traffic that business owners need to consider. Automated traffic can chew up bandwidth and impact analytics without generating any revenue. That could result in higher than expected bills, which could hurt the bottom line of companies that have smaller infrastructure budgets. While customer acquisition is expensive, human visitors are more likely to result in a sale, a subscription, or a viewing of content. Additionally, as bot visitors increase, it becomes more difficult for business owners to connect web traffic with customer demand. Garg suggests that the era of winning customers with creative web design is coming to a close, and future online iterations should move away from visual interfaces and more toward lean sites that emphasize behind-the-scenes data exchanges. “The internet is shifting from a destination [that] humans visit to an invisible infrastructure that AI agents navigate for us,” he says. “Your tech budget needs to shift from beautiful UIs to robust bot interfaces. Don’t build a better website. Build an MCP (Model Context Protocol) server that lets the AI ecosystem seamlessly transact with your business.” For instance, one SaaS company, Monday.com, has created an AI agent-only sign-up flow on its website. It employs a reverse CAPTCHA system that it says only AI can get through. That could increase your business’s chances of AI chatbots recommending your site to users, much like today’s search engines do. The shift from primarily human users to primarily bots is one experts have been predicting for quite some time. Automated traffic across the internet grew almost eight times faster than human activity in 2025, according to the 2026 State of AI Traffic report from cybersecurity firm Human Security. Cloudflare calls it the next phase of the internet’s evolution, but cautions that it will create challenges that current IT infrastructure and cybersecurity were not designed to handle. “IT leaders now face fundamental questions about trust, visibility, and control that traditional architectures can’t answer,” the company said in a blog post. “The organizations that recognize this shift — and redesign their infrastructure accordingly — will shape how the internet evolves. Those that don’t will find themselves constantly outmaneuvered.” BY CHRIS MORRIS @MORRISATLARGE

The Flaws in Mass Layoffs for AI Productivity Are Beyond Obvious Now

Just when I thought I was done calling out tech CEOs for horrible mass layoff decisions, one of those CEOs doubles down on the mass layoff rhetoric. So here’s what we’re gonna do. You know how, when your mom or dad tried to give you solid life advice that just didn’t stick, they ended up sitting you down and listing out the flaws in your reasoning one by one? Well, sit down, kid. And like I tell all my kids: Look, I’m gonna yell at all of you, but I’m really only yelling at one of you. You’ll quickly figure out if you’re the one I’m actually yelling at, but it’s not gonna hurt any of you to hear what I have to say. This Could Be Any Corporate Tech CEO My opening histrionics aside, I want to make it clear that I’m not really trying to smack anyone individually. What I’m about to criticize is not the work of a single misguided leader, it’s the culmination of a spreading misguided follow-on leadership strategy. I also want to apologize if any of this comes off as flaming the writer or the publication behind the article I’m going to use as an example, because it literally could have been any tech CEO speaking to any publication after cutting double-digit percentages of their workforce and being all super-pumped about the future. It is such a bad look, so good job exposing it. But I guess we’re all numb to it now, when we read things like: “What [tech company’s] mass layoff tells us about the future of work” I encourage you to read the article. Go ahead and give the author some respectful clicks, because they get it right at the end, with facts. But ultimately you don’t have to read it because I’m about to take apart the “fire-all-the-humans-and-replace-them-with-AI” strategy point by point. This Isn’t a New or Novel Strategy Let’s start at the top: “Zeb Evans, CEO of the collaboration software startup ClickUp, claims that this shift is imminent. Last Thursday, Evans announced on X that the company, which was last valued in 2021 at $4 billion, had laid off 22% of its workforce.” A couple of things are quickly evident. First, this appears to be the same “everybody’s gonna eventually do it” reasoning that Jack Dorsey used when Block laid off 4,000 people just a couple months ago. Second, it’s not a coincidence that the company was last valued in 2021 at $4 billion or that that might have been its peak. 2021 was the acceleration point of the Great Labor Arms Race, and Corporate Tech companies across the industry started hiring people ill-equipped for poorly-defined roles at salaries that would have broken the bank if money wasn’t so cheap—or even free, via automatically forgiven loans. But the pretzel logic that gets used here is what makes this follow-on strategy especially obtuse. When Cutting Costs Isn’t Cost-Cutting The CEO “characterized that reduction as not a cost-cutting measure, but rather a radical embrace of AI that will propel the company to the next level.” Quick question: When is cutting 22 percent of your labor costs not a cost-cutting measure? I’ll just let that one hang. I’m not a mean person, really. The Biggest Mistake a CEO Can Make “‘Most savings from this change will flow directly back into the people who stay. We’ll be introducing million-dollar salary bands. If you create outsized impact using AI, you’ll be paid outside of traditional bands,’ Evans wrote.” One of the best lessons a mentor ever gave me about leadership is: The worst mistake a leader can make is looking at a chart that goes up and to the right and believing that chart will always go up and to the right. It never works out that way, but the temptation to think it will is always there. Using my mentor’s advice, I have several questions: Is this CEO talking about paying a percentage of profits based on whatever metrics they invent to show “outsized impact created by using AI”? Does the CEO plan to keep paying that employee their $1-million salary when the “outsized impact created using AI” returns to the mean? Quick follow up: Does that keep going until it breaks the budget or is this just, like, an MLM thing? If not, and the CEO does the sensible thing that every other company in the history of companies has called “commission,” won’t that employee just hop to the next company when that company offers them a $1-million salary to do what they just did? And then finally, let’s do the cut-throat AI thing that serves as the reason for the 22 percent “savings” in human labor: Once that “outsized impact created by using AI” materializes, why do you still need that employee? Especially if you’re now paying them a million-dollar salary? Isn’t that more “savings” just waiting to be “saved”? Making a Fortune Babysitting That last question kind of introduces another question: What are we paying these employees a $1-million salary for? “ClickUp recently introduced roughly 3,000 internal AI agents to handle a wide range of complex tasks on behalf of its employees…. Instead of performing the work themselves, staff members are now expected to direct these agents and ultimately review the output to ensure it meets the company’s standards.” Are we planning on paying million-dollar salaries to babysit agents? Because the last year has shown us that’s not where the seven-and-eight figure salaries are going. The 100x Productivity Myth “Evans’s goal, according to his X post, is for AI to turbocharge ClickUp into a ‘100x org.’” I’ve been on the AI front for over 16 years, and I get called a “100x guy” or a “10x guy” a lot. I don’t know what that means, but it sounds cool so I just smile and say thank you and get back to the data. Actually, I do know what it means, in another context, because I’ve spent my entire career as an entrepreneur and/or consultant, and have worked with a vast array of venture capital and private equity firms and their strategies. One of those strategies, familiar to everyone, is to create ROI by what I’m going to dub “numerator-maxxing” (see, I can make up buzzwords too). The strategy starts out logical enough. The company that the firm is investing in is doing something very right. Their numerator—the value that the company is generating—is a lot higher than their denominator—the money and sweat effort and brainpower being put into the company. The firm believes that the company’s numerator is artificially low and is being constrained by a weak denominator. So the firm dumps a bunch of money into the denominator. That’s their bet. When this happens, the numerator almost always increases. Where it goes wrong is when, a couple years into it, the numerator has not increased by orders of magnitude to compensate for being weighted down by a heavy denominator. One hundred over 10 is a much bigger number than 1,000 over 1,000,000. Sorry for the math. So yeah. Machines are less weight in the denominator than humans, and you can also add exponentially more of them to the denominator without it getting much heavier. But what do they add to the freaking numerator? Where AI and agent productivity is concerned, no one—no one—is looking at the numerator. Well, no, wait. Gartner took a look. Vindication Isn’t What It Used to Be Like I said, the writer gets a lot right at the end, and does it without my cartoonish fist shaking. In response to the metric most commonly used to measure that “outsized AI impact”: “[C]ritics argue that “tokenmaxxing”—as this concept is known—is the wrong metric because it simply racks up AI expenses.” In response to the company’s incredibly circular claim that people who automate their jobs with AI will always have a job: “But if AI keeps taking over more tasks, ClickUp will eventually need fewer and fewer people.” However, the most damning truth came from a quick mention of a quiet study from Gartner on ROI from AI-as-labor-replacement, just three weeks ago, from which I wish the writer had pull-quoted: “Many CEOs turn to layoffs to demonstrate quick AI returns; however, this disposition is misplaced,” said Helen Poitevin, Distinguished VP Analyst at Gartner. “Workforce reductions may create budget room, but they do not create return. Organizations that improve ROI are not those that eliminate the need for people, but those that amplify them by aggressively investing more in skills, roles and operating models that allow humans to guide and scale autonomous systems.” Am I still Don Quixote for screaming about this for the last 16 years? In fact, I said the same thing yesterday when highlighting the unintended consequences of this misguided follow-on leadership strategy, and it still feels like the loud part being said far too quietly. It will always be the humans behind the tech that will make the tech successful—not the babysitters, not gamification, not agentmaxxing, tokenmaxxing, or numerator-maxxing. So if any of those leaders see this rhetoric and still believe these mass layoffs are about AI and not about mistakes made by leadership a few years ago in hiring the wrong people for roles that were never clearly defined at salaries that never should have been offered, I’m begging you, think twice before you agentmaxx. The flaws are now obvious and documented. There is nowhere left to hide. EXPERT OPINION BY JOE PROCOPIO, FOUNDER, JOEPROCOPIO.COM @JPROCO

Anthropic suspends all access to Mythos model after US government bans foreign nationals use

AI company Anthropic has disabled customer access to its most capable systems after the US government ordered it to suspend all use by foreign nationals, Anthropic said in a statement Friday evening. The move is the latest in a series of adverse Trump administration actions targeting the company. The broad directive to Anthropic’s Mythos 5 and Fable 5 models is one of the furthest-reaching actions the government has taken in response to the advanced capabilities of an AI model. Anthropic said the US government gave it the directive, citing “national security” issues. The company said the government didn’t provide specific details about the national security concerns, though it believed the government had “become aware” of a method of “jailbreaking” Fable 5, or getting around its internal safety guardrails. “We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities,” Anthropic said in its statement. “These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass.” Anthropic said it had instituted several safeguards for its newest models to “greatly reduce the likelihood” that they are “misused for tasks related to cybersecurity,” noting they’ve received complaints from users about those guardrails being too strict. Anthropic also noted it has worked with the US government to “red team” Fable’s safeguards and that no model is completely resistant to any jailbreak. Anthropic said that while they are complying with the directive and removing access to the models for everyone, “we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people.” “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers,” the company added. The restriction means that many foreign nationals working for Anthropic will not be able to touch those models. The Commerce Department, which issued the restriction, did not immediately respond to a request for comment. Axios reported the government’s directive would require Anthropic to obtain a license “for the export, re-export or domestic transfer of those Anthropic models.” Anthropic’s newest model, Mythos, has spooked the US government and Wall Street with its capabilities, which experts say can exploit cybersecurity vulnerabilities at an unprecedented pace. The model was seen as so capable, Anthropic initially limited its release to a group of key partners in order “to secure the world’s most critical software.” Anthropic released Fable 5 last week as a version of Mythos that is safe for general use. The model also helped spark the Trump administration’s recent executive order on AI, which asks companies to voluntarily share new models deemed to have advanced cyber capabilities with the government up to 30 days before providing access to other partners. One source with knowledge of early discussions of the executive order said the idea of banning foreign nationals from working on such models had been floated for that order, but the idea never made it into a draft. The government has had a complicated relationship with Anthropic. Earlier this year, the Trump administration blacklisted the company, declaring it a “supply chain risk” in military dealings over Anthropic’s insistence that the Pentagon include certain safety guardrails for the government’s use of AI in warfare. Anthropic sued the government over the designation as “unprecedented and unlawful” and notched at least one early win in the ongoing case. Despite President Donald Trump’s directive at the time of the designation for all of the federal government to cease working with Anthropic products, the White House has stayed in close touch with the company, and some parts of the federal government have found a workaround to continue accessing Anthropic’s models, especially after the release of Mythos. Anthropic was also deeply involved in helping draft the latest executive order, sources familiar with the situation told CNN, and its executives had been invited to the White House for a signing ceremony that was ultimately canceled at the last minute. By Hadas Gold

AI Is Wreaking Havoc at Starbucks and Pizza Hut. Social Media Is Having a Field Day

AI woes are coming for the food service industry, and social media can’t help but celebrate. This week, both Starbucks and Pizza Hut made headlines for negative news about their internal applications of artificial intelligence. At Starbucks, an inventory tool got the chop after making frequent counting mistakes, while at Pizza Hut, a delivery tool drove a franchisee to file a lawsuit. Social media users are saying the two stories may point to a larger trend—that for the first time in the AI era, more companies will pull away from AI than embrace it. Starbucks walks back an AI tool On Monday, Starbucks told employees it was retiring an inventory-counting tool powered by AI after the technology led to inaccurate counts and mislabeled products. “Starting today, Automated Counting will be retired,” said an internal company newsletter verified by Reuters. “Beverage components and milk will now be counted the same way you count other inventory categories in your coffeehouse.” In a statement to Fast Company, a spokesperson for Starbucks explained that the company’s choice to axe its Automated Counting tool is in line with its larger AI strategy, which is based on trial and error. “We test ideas in our coffeehouses, listen closely to partner feedback, and make changes to deliver a better, more consistent experience.” Starbucks’s move to ditch one AI tool doesn’t mean the company is forgoing the technology entirely. The company is still investing in internal AI applications, including an AI assistant for baristas called Green Dot Assist and an AI-powered order-sequencing system called Smart Queue. The brand is also experimenting with an integrated Starbucks app within ChatGPT. Pizza Hut’s delivery system backfires Where Starbucks’s choice to nix its AI tool came from the top down, the anti-AI sentiment at Pizza Hut started with a disgruntled franchisee. In a lawsuit filed on May 6, franchisee Chaac Pizza Northeast, which operates more than 100 Pizza Hut locations, alleged that the company forced it to adopt an AI tool called Dragontail, which inadvertently pushed average wait times from under 30 minutes to over 45 minutes in more than half of all orders. The complaint explained that the issue wasn’t with Dragontail itself, but with the information the tool provided to DoorDash drivers. Dragontail is meant to optimize food delivery by giving delivery drivers real-time updates on order preparations and timing. But according to the lawsuit, its implementation in 2024 caused “cascading operational breakdowns and customer dissatisfaction,” resulting in more than an estimated $100 million in lost business and enterprise value. Reportedly, once DoorDash drivers could see the real-time status of multiple orders through Dragontail, they would wait inside restaurants until multiple orders were ready, meaning some orders were being held for up to 15 minutes after they were ready for delivery. Because Chaac Pizza Northeast relies on DoorDash for all of its deliveries, the franchisee alleged that the forced change to its delivery model had a major impact on its sales. At its New York City locations, Chaac said its sales swung from positive 10.19% to negative 9.78% after implementing Dragontail. “With the intention to improve efficiency and service to the customer, Dragontail did the exact opposite,” the lawsuit stated. “It caused significant delays and pummeled consumer satisfaction.” Pizza Hut has not responded to Fast Company’s request for comment. Social media sees a trend With the stories from Starbucks and Pizza Hut breaking in quick succession, social media users are drawing connections between the two food service chains’ AI troubles. “Over the next 1-2 years we’re going to start hearing more reports about companies pulling back from AI than adopting AI, and markets aren’t ready,” one X user theorized. “The AI bubble might burst quicker than I thought,” echoed another. “You’re going to be hearing a lot more about forced AI integration and what a disaster it is for businesses and consumers,” a third user agreed. Other users pointed out that all of these problems could have been avoided if tasks hadn’t incorporated AI in the first place. “To err is human,” one person quipped, “but to really screw things up, you need a computer.” By Jude Cramer

Instagram’s AI Support Bot Made a Costly Mistake. It’s a Warning for Every Company

Sometime last weekend, hackers asked Meta’s AI support bot for access to someone else’s Instagram account. The bot said yes. The method, reported first by 404 Media, was almost insultingly straightforward: instruct the chatbot to add a new contact email to a victim’s account, confirm the change with a code sent to that email, then use it to reset the password. In some cases, all that was also needed was a VPN to spoof the victim’s location. Among the accounts taken over: the former White House page from Barack Obamaʼs presidency, cosmetic chain Sephora, and the chief master sergeant of the U.S. Space Force. While this hack was obviously a security flaw, most businesses should consider that this blunder also reveals a management error. The front door is now automated AI customer service is everywhere. According to MarketsandMarkets, the global AI customer service market was valued at $12.06 billion in 2024 and is projected to reach $47.82 billion by 2030, growing at a compound annual growth rate of around 25.8 percent. The economics are hard to argue with: AI scales cheaply, doesn’t require shift management, and doesn’t call in sick. AI has been shown to cut response times by 37 percent and reduce operational costs by 35 percent. Companies have moved quickly. Banks use it for dispute resolution, airlines use it for rebooking, telecoms route complaints through it, and e-commerce platforms use it to handle returns. Around 80 percent of companies are either currently using or planning to adopt AI-powered chatbots for customer service by 2025. Meta is not unusual for deploying AI support at scale. When Meta launched its AI support assistant in March, it was presented as a way to help users reset passwords, regain account access, and report problematic content. These are not trivial tasks. Account recovery involves identity, security, and trust—the kind of territory that, not long ago, companies deliberately kept behind a human review process because the consequences of getting it wrong were considered too serious to automate away. Meta has also reassigned thousands of employees into AI-related roles and used AI to automate risk assessments of updates, safety features, and content moderation—all of which are tasks that previously required human review. Therefore, the support bot exploit arrived in an organization that had been replacing human judgment with automated processes across multiple functions simultaneously. Authority without judgement The exploit worked not because the AI behaved unpredictably, but because the AI behaved more or less as it was designed. A user asked it to update a contact email, and it updated the contact email. The system executed a process it was authorized to execute, but in doing so, handed a hacker the keys. This distinction matters enormously for anyone responsible for deploying AI systems. Authority can be automated, but judgment is a different problem entirely. A junior employee in a similar situation might also make a bad call. Humans fail too, but organizations know how to supervise, train, audit, and escalate human decisions. Real people typically offer something an AI system can’t: the ability to notice that something feels off. In this case, hesitation at an unusual request, the instinct to double-check account history, and the recognition that a request arriving via VPN from an unfamiliar location (asking to change contact details on a high-profile account), warrants a pause that could have prevented disaster. Gartner forecasted that 20 to 30 percent of businesses will replace human customer service agents with AI by 2026. The efficiency case for doing so is well established, but what’s less well examined is the category of decisions that look like processes but aren’t. Account recovery looks like a process, but embedded within that process is a judgment call: Is this request legitimate? Humans often make that call poorly too. The difference is that when a human makes it poorly, the failure is legible. There’s accountability. There’s someone to fire. Victims of the Instagram account takeovers told 404 Media there was no way to escalate their problem and speak to an actual human. Historically, high-stakes support requests — account recovery, fraud disputes, identity verification — were escalated to human agents because the potential for harm was too high to leave it to a scripted response. The escalation path was a precaution for the edge cases that automated systems weren’t designed to handle. Research published in the Journal of Consumer Research found that customers evaluate service provided by bots less favorably than identical service provided by humans, partly because they perceive automation as the firm cutting costs at the customer’s expense. That perception hardens quickly when customers discover there’s no human available at all. The harder problem The lesson here isn’t that companies shouldn’t deploy AI in customer service, but the quality of the decisions being made around where AI gets authority and what happens when it exercises that authority incorrectly. The questions that matter aren’t technical; they’re structural. What decisions can AI make autonomously, and which ones require a human to sign off? What triggers an escalation, and who owns it? When the AI gets something wrong — not if, but when — can a customer reach someone who can fix it? Gartner concluded in mid-2025 that half of the organizations expecting to significantly reduce customer service headcount because of AI would abandon those plans by 2027, and that 95 percent of service leaders planned to retain human agents in a digital-first but not digital-only model. The companies that arrived at that conclusion before Meta’s incident are in a better position than the ones who’ll arrive at it after. The Instagram story will probably be remembered as an AI security vulnerability, and will likely get fixed as one. As of June 3, it reportedly still hadn’t been patched. As companies race to automate customer interactions in pursuit of lower costs and faster resolution times, they’re discovering that it’s far easier to reduce the cost of executing a decision than it is to replace the judgment behind it. With better training, retrieval-augmented verification, anomaly detection for suspicious requests, and well-designed human-in-the-loop escalation, AI systems may be able to take on increasingly sophisticated judgment-like tasks over time. But deciding which requests deserve more scrutiny than the process allows remains, stubbornly, a human problem for now. AI can execute authority at scale, but the question that the Instagram incident forces into the open is a simpler one: When it gets things wrong, who’s responsible? Right now, the answer at too many companies seems to be nobody. BY CONNOR JEWISS

How to AI-proof your job

Meta. Nike. Intuit. UPS. It seems every day a new company announces layoffs, often citing AI as the cause. AI is already reducing US monthly payroll growth by roughly 16,000 jobs in the past year, according to a recent Goldman Sachs report. Knowledge workers face the sharpest exposure, as their output is exactly what AI replicates best, at superhuman speed, around the clock. “The most valuable jobs, the ones that we tell people to go to school for – software engineer, finance professional, accountant, lawyer – a lot of these cognitive professions, those are the ones that are the most vulnerable… to AI automation,” David Shrier, professor of AI & Innovation at Imperial College London, told CNN. But humans will always be needed in some form – and there are steps you can take to increase the chances you’ll protect your own job. Audit your role Before you can future-proof your career, you need a clear-eyed view of what you do in your job. Think of jobs as a “collection of tasks we switch between, often many times a day,” Oded Nov, a professor of technology management at New York University, told CNN. Consider which of those tasks are the most repeatable, rule-based computer tasks, like processing expense reports, which takes raw data and converts it into a different form. The more predictable a function, the more vulnerable it is to automation. The CEO of Cloudflare wrote in an opinion piece for the Wall Street Journal that he recently laid off 20% of his work force, focusing on “measurers”: middle management and those who work on audits, operations, compliance, etc. “AI isn’t coming for builders or sellers, but it is coming for measurers,” he wrote. “Tireless, independent, efficient and available, AI systems can now measure an organization with a level of objective detail and precision that was previously impossible even for the best employees.” Some jobs, such as those in hospitality, healthcare and skilled trades, still need someone physically present to do much of the work. Robotics is at least a decade away from replacing those roles. Invest in skills that are structurally hard to automate After your self-audit, focus on skills that are not repeatable, predictable and rule based. In addition to physical duties, AI is not yet as good at handling tasks that require emotional and social awareness, such as “understanding organizational culture or group dynamics,” Nov said. AI tends to be recursive, rather than inventive or creative. “AI is bad at creativity, but it’s surprisingly good at elaborating on creative prompts,” Shrier said. “But you still need the human to come up with the idea and guide the AI to do something interesting.” Invest in those skills. If part of your job involves sales and convincing people to sign a contract, focus on the interpersonal skills that help you build trust with clients. Customers might go to AI to research, but they usually still want to deal with a human being when making big purchases. Embrace AI and learn how to make it work for you AI will soon become a pervasive part of our lives, just like the internet. Get familiar with the major AI systems – ask various AI chatbots about your job and how they can help make your work more efficient, then give their suggestions a try. Play around with new coding tools that help you create your own app and website without needing to write your own code. But chatbots aren’t what makes AI so useful in the workplace, it’s AI agents, or programs that run automatically, make decisions and take actions autonomously. You can learn how to make your own, and a chatbot can help. Try the prompt: “I want to learn how to make an AI agent. Walk me through the steps to create an agent that can do [specific task]”. “In some ways it’s never been a better time to be an entrepreneur, because if you can think of it, you can make it,” Shrier said. “There are people making robust enterprise-grade software that is built off of a prompt in plain English.” Humans can’t be entirely replaced Even roles that include a lot of AI-achievable tasks will still need humans in some way. AI has particularly affected the coding industry. But Anthropic employees still edit and review code even if they’re not writing it, CEO Dario Amodei said at the World Economic Forum in January. “In best-case scenarios, the more mundane tasks that are part of people’s jobs today will be handed over to AI, while the more interesting and rewarding tasks will be done by humans, probably with some support of AI,” Nov said. “It’s very likely, if history is a guide, that new jobs, consisting of new collections of tasks, will be created.” By Hadas Gold

AI ‘voice cloning’ scams are on the rise. Here’s how to protect yourself

A California mom says she was scammed out of thousands of dollars this month after receiving a call that sounded like her daughter in distress. She now suspects it was an artificial intelligence-generated hoax. She’s one of many who have been targeted by so-called “voice cloning” scams as AI tools allow anyone to create a convincing replica of someone’s voice with only a few seconds of real audio. Americans lost more than $893 million to AI-related scams last year, including voice cloning attacks along with AI-generated phishing emails, romance scams and other hoaxes, according to the Federal Bureau of Investigation. Scammers can mimic anyone from family members and friends to coworkers or professional services workers. Banks including the United Kingdom’s Starling and the Commonwealth Bank of Australia have warned customers to watch out for voice cloning scams. Experts say AI voice replicas have gotten so realistic that most people can no longer reliably distinguish them from real human voices. “For the everyday person, it is just not fair to expect them to be able to spot this stuff,” said Henry Ajder, an expert on AI-generated media who consults for governments and companies. “I struggle with it. Most people do.” How do AI voice scams work? Scammers can create an AI replica of someone’s voice using a short recording of their speech — often pulled from social media or an earlier scam call that was surreptitiously recorded. Social media can also provide a trove of information about family members and close friends who could be targeted. Fraudsters will typically make it sound like the loved one they’re mimicking is in distress, purportedly having been kidnapped or in jail. Then they’ll urgently demand money in exchange for their loved one’s release. “There was no time to think,” Gary Schildhorn, a Philadelphia attorney who was targeted by an AI voice scam mimicking his son, told CNN last year. “It was all, ‘I have to react to help my son. He’s in trouble.’” In some cases, the AI voice may be more than just a single recording. Sophisticated attackers could use text-to-speech tools or “voice skinning,” which manipulate a scammer’s voice so they sound like the person they’re imitating in real time. Those techniques facilitate back-and-forth conversations between the target and the AI clone voice, potentially making the scam more convincing, Ajder said. Hackers can also make it appear as if a call is coming from a known number through a tactic known as caller ID spoofing — so you can’t necessarily trust that a call that appears to be coming from your mom is indeed her. How to avoid falling victim to AI voice scams Strange pauses or vocal fluctuations were previously considered red flags that a caller’s voice might be AI-generated. But those signals may no longer be present now that AI has advanced. Instead of trying to determine whether a voice is authentic, look for other general scam warning signs, Hany Farid, a professor at UC Berkeley and chief science officer at GetReal Security, told CNN last year. Is the person on the other end giving a deadline or introducing a sense of urgency? Are they encouraging you not to tell anyone else what’s happening? Are they asking you to move large sums of money in unusual ways? Those are the types of questions experts say to keep in mind. Targets who receive these types of calls should try contacting their loved one through other means, such as via a text message, calling them on another person’s phone or reaching out to someone who would know where they are. Deborah Del Mastro, the California mom recently targeted, told ABC7 News that she called her daughter only after sending money to the scammers. Her daughter answered right away and was at work. Families or coworkers can also establish a precautionary “code word” that can be used to verify each other’s identity. It should be a word or phrase that only a small group of people know and isn’t discoverable online. “Ultimately, if you suspect that something might not be right, it is much better to have your mum or your brother or your friend laugh at you for thinking that they’re a robot,” Ajder said, “than it is to potentially be running to an ATM.” By Clare Duffy

Nobel Prize-Winner Demis Hassabis Says AI Job Cuts Are Dumb. Research Agrees

Across America, graduating college students are booing commencement speakers who mention AI, and communities are battling data centers. A backlash against AI is clearly underway. It’s not hard to see why. With their talk of a jobs apocalypse and general creepiness, plenty of AI CEOs these days don’t seem to realize they often sound like Hollywood supervillains. But there is one notable exception — Demis Hassabis. If you’re not already familiar with him, Hassabis is the head of Google-owned DeepMind. That seems like a position with supervillain potential. But he also won a Nobel Prize in chemistry for his work using AI to solve one of biology’s most vexing problems — predicting the shapes of proteins — and then releasing the solution to the world’s scientists for free. Rather than using AI to replace workers, cheat on everything, or flood the world with slop, Hassabis enthuses about its potential to help cure disease and engineer energy abundance. He’s soft-spoken and bespectacled rather than square-jawed and cape bedecked, and he positions himself as AI’s science-minded good guy. His latest public stance is likely to add to this reputation. As Wired recently reported, he called the recent mania for AI-fueled job cuts basically just dumb. Why AI-driven layoffs are dumb Speaking at Google’s I/O event, Hassabis explained that DeepMind doesn’t look at AI as a way to eliminate jobs and cut costs. Instead, he sees the technology as a way to help businesses dream bigger and do more. “Perhaps there is an ulterior motive for putting those messages out; raising money or whatever,” Hassabis said of other firms slashing headcount supposedly because of AI. “From my point of view, from DeepMind and Google’s point of view, if engineers are becoming three or four times more productive, then we just [want to] do three or four times more stuff.” Sure, companies can use potential productivity gains of AI to trim headcount and pad profits for the benefit of investors. But there’s no shortage of other ways they could deploy the resources freed up by handing many tasks over to the bots. “I have a million ideas, from lab drug discovery to game design,” Hassabis continued. “I’d love to have some free engineers to go and do those kinds of things.” Companies trimming tech talent may be suffering from “a lack of imagination—and a lack of understanding of what’s really going to happen,” he believes. Research shows that layoffs usually backfire Is Hassabis right that companies chasing a short-term boost through AI job cuts may come to regret the decision? We’ll all have to wait for a definitive answer to that question. But there are good reasons to think he might have a point. First, research shows that layoffs in general are often a stupid idea. Of course, they save money. Sometimes they’re genuinely unavoidable (though that mostly doesn’t apply to the mostly super profitable firms doing them now). But multiple studies show that a short-term bump in performance is usually followed by a longer-term decline. Stanford’s Jeffrey Pfeffer summed up the research this way: “Layoffs often do not cut costs. … Layoffs often do not increase stock prices, in part because layoffs can signal that a company is having difficulty. Layoffs do not increase productivity. Layoffs do not solve what is often the underlying problem, which is often an ineffective strategy, a loss of market share, or too little revenue. Layoffs are basically a bad decision.” That’s not even taking into account the human cost to workers. Other research has found that being laid off increases a person’s risk of suicide by two and a half times and increases mortality by 15 to 20 percent over the following 20 years. So why do layoffs? As Hassabis suggests, some firms may be peakcocking for Wall Street. Others may genuinely need to free up funds for a strategic shift. But mostly, Pfeffer insists, “layoffs are the result of imitative behavior. These companies … are doing it because other companies are doing it.” An AI-specific reason layoffs are dumb That’s the general case against layoffs, and it’s pretty strong. But there is another argument against AI-driven layoffs specifically. On the MIT Sloan Management Review website, consultant and author Andrew Winston points out that when firms use AI as an excuse to trim headcount, they also diminish their pipeline of talent. They may save a few dollars. They may also find themselves short of skills and talent when they need them in the future. You can ignore this problem, but it might come back to bite you. As the case of climate change proves, asking CEOs to forego a quick performance bump for longevity and resilience later may be borderline hopeless. Winston continues to point out this inconvenient truth about AI anyway. “I’m asking companies to accept potential (short-term) competitive disadvantages on the basis of uncertain future benefits and collective responsibility. That’s a hard sell,” he acknowledges. “But I have a sneaking suspicion that we will look back at early 2026 and kind of wish we had just stopped.” Don’t get caught up in AI layoff FOMO All of which is a warning to other bosses looking on at tech company layoffs with FOMO. When some of the biggest names in business are all slashing jobs, you might feel like you’re missing a trick if you don’t follow suit. But both Nobel-winner Demis Hassabis and a whole bunch of research suggest leaders should take a breath and really consider whether they want to get swept up in AI job-cut mania. Cutting costs has obvious benefits for your bottom line. But what profit-making possibilities will you not pursue because of those job losses? How will the departure of their colleagues affect the state of mind and productivity of those left behind? Will you likely regret the move later when you find your team is burned out and your bench of experienced talent is incredibly thin? Before you cut jobs, take a minute and consider whether Hassabis is right. Doing more things because of AI could be a smarter choice than employing fewer people. EXPERT OPINION BY JESSICA STILLMAN, CONTRIBUTOR, @ENTRYLEVELREBEL

AI May Replace 80 Percent of Skills. This Last 20 Percent Will Make You Irreplaceable

I work in front of a screen. And I’ve been thinking about how AI will change my work. What does it even mean for my future? It’s completely normal to wonder about this. Most people are convinced artificial intelligence is a threat to their careers. But what they are forgetting is the human value they bring to their work. Aaron Levie, CEO of the enterprise cloud company Box, recently pointed out that when people watch AI at work, they are most likely seeing it take over the first 80 percent of a task—the heavy lifting of repetitive processing. The last 20 percent is where you come in. Your domain expertise, judgment, and relationships. That is what makes you irreplaceable. AI can finally give you the space to add human value at work. “The extra 20 percent, it turns out, is all the value creation of that profession. All the expertise and domain knowledge is in that last 20 percent, not the text that got generated,” Levie said in an interview with Casey Newton of Platformer, the online publication about tech and democracy. I couldn’t agree more. Your judgment is valuable Take the work of a lawyer. Junior associates spend most of their week reading precedents, looking for case connections, and summarizing legal statements. That’s the 80 percent of the work. The long, tedious, trainable, reproducible task. No client hires a lawyer just for that. They expect them to make a better and more persuasive case for them to win. To convince the judge. To save the dying deal. The 20 percent only you can do. The practical human value. AI work feels like completion, but it’s not. Not even close. It’s good at execution, but the meaning and context are all up to you. The career anxiety you feel about AI is normal, but it may be misdirected. When people say “AI is taking my job,” they usually mean it’s taking their tasks. Writing code, analyzing long documents, and doing the research. The first pass. And yes, super-intelligent machines are coming for those. If you built your professional identity entirely by executing tasks, that’s hard reckoning. The good news is, your knowledge from doing and experience is still relevant. All of that makes your judgment valuable. AI cannot replicate that. Domain expertise under pressure must count for something. A cybersecurity engineer knows exactly what steps to take when an attack is live. Making that call in real time with incomplete information changes their approach. Data doesn’t always give a clear answer. You have to decide anyway. Who bears that decision? Not the AI. It can notice the patterns. But it’s the engineer who must come up with a specific solution to solve the problem. Most companies are limited by execution capacity. They can’t pursue every good idea because they don’t have the people to execute them all. When AI takes care of the execution, the constraint becomes the quality of the ideas. The clarity of human judgment. And the client relationships they can’t afford to lose. If you still want to hold onto the 80 percent, you’re running in the wrong direction. You can’t compete with AI on speed. Focus on honing your quality of judgment. The value and usefulness only a human perspective can fill. Your clients don’t buy your services just for the deliverables; they buy the peace of mind too. And they also like to work with people with a better reputation. Trust is not a digital asset. In the future of work, the world will reward the 20 percent more. The calculator case When calculators became universal, they didn’t make mathematicians obsolete. They just took over repetitive math, which freed mathematicians to spend more time on quality and better mathematical thinking. The profession evolved upward. The entry-level work disappeared. The high-level work expanded. AI is doing the same thing to knowledge work. At scale. Only it’s happening everywhere, in every field, all at once. Build the kind of expertise that requires better judgment. And relationship capital that compounds over time. Develop your specific point of view at work. What AI can’t do is replace your original perspective earned through engagement with practical problems over time. Your distinct angle on your field, built from specific experience, failures, and observations is what matters. It’s the experience that makes your presence valuable in the room. The threat from AI is understandable. The pace of change is insane. Some jobs and skills are becoming less valuable. Don’t stay terrified. Fear takes away your ability to think clearly. It makes us hold onto the routine tasks we feel safe doing. But routine tasks are exactly what the machines want. If you’ve been doing meaningful work for any serious amount of time, you’ve accumulated things AI cannot access. AI is taking the parts of your job you probably didn’t love that much anyway. Even that requires your input. If you feed AI the wrong ideas, it will give you a brilliant, highly optimized wrong answer. The rarest skill right now is the ability to diagnose the actual problem before rushing to fix it. You are still needed for work that requires your specific experience. Don’t underestimate what you’ve already built. You have what it takes to survive AI. By Thomas Oppong FAST COMPANY