The FBI has issued an urgent warning about a fast-acting phishing scam operated by the hacking platform Kali365.

The scam targets Microsoft 365 users — anyone using Teams, Outlook, or OneDrive could be in danger.

The goal of the phishing scam is to trick potential victims into handing over their OAuth device codes, which the hackers can then use to bypass multifactor authentication and access Microsoft accounts. To stay safe, you’ll have to keep your OAuth tokens to yourself.

What Does the Scam Look Like?

If you’re targeted by Kali365, you’ll receive a phishing email that appears to be from an authentic document-sharing platform. It will have a device code and instructions on how to verify yourself.

The FBI’s warning, available online here, walks viewers through the four steps of the scam:

• Lure: An email “impersonating trusted cloud productivity and document-sharing services,” containing a code, which asks users to go to “a legitimate Microsoft verification page and enter the code.”
• Authorization: Entering this device code will actually give the scammer’s device access to the victim’s account.
• Token Theft: The scammer then “captures OAuth access and refresh tokens, granting them access to the targeted individuals/entities’ Microsoft 365 account.”
• Persistence: Armed with this access, the scammer can now gain access to even more Microsoft 365 services, like Outlook, Teams, and OneDrive, without needing to get any additional passwords or information.

What’s Kali365?

The threat doesn’t come from any single, unified hacking group. Instead, these phishing attempts might come from any scammer who own the Kali365 platform, a subscription-based service aimed at helping scammers become more effective.

News of this “scamming-as-a-service” software tool first emerged to the general public in April 2026, but has been circulating through closed groups on Telegram for a lot longer. According to a Bitdefender report, it costs about “$250 per month or $2,000 a year.”

How Can You Stay Safe?

The biggest thing you can do to avoid this particular scam is stay vigilant against any phishing emails. However, that’s easier said than done.

The FBI’s warning comes with a few technical tips for staying safe.

You can try to “create a conditional access policy to block device code flow for all users,” and you can figure out how to block authentication transfer policies, so that any users with unauthorized access can’t transfer their authentication between computers and mobile devices.

Finally, if you’ve already been hit, you can report the incident and help experts work to stop the evolving threat: Go file an online complaint with the Internet Crime Complaint Center (IC3) here if you’ve been impacted by the Kali365 phishing kit.

The companies most exposed to AI achieved labor productivity gains of 163% in a new study, significantly leapfrogging other businesses.

Within the labor market, AI is creating two different kinds of roles. “Professionalized” roles, where AI acts more as a force multiplier, are seeing a greater growth in hiring and wages compared to “democratized” roles, where AI makes a role easier to perform.

Companies are changing the requirements of roles as AI continues to sweep the workplace. This is most evident in entry-level roles, where human skills such as judgement and creativity are becoming more desirable.

The Companies Most Able to Use AI Are Pulling Ahead

Companies most exposed to AI, dubbed “super-star companies,” have seen labor productivity growth of 163% relative to 2018, in the 2026 PwC Global AI Jobs Barometer. These companies are in the top 20% of the most AI-exposed companies overall.

Companies operating within the most AI-exposed sectors recorded a 34% growth in productivity in 2025 compared to 2018, in contrast to the 24% recorded for the companies least able to use AI.

Likewise, headcount growth at the most AI-exposed companies is higher than the least exposed companies. Companies with more exposure saw 52% growth, compared to the 36% seen by those with less exposure.

AI is Creating Two Types of Roles

PwC reports that AI is creating a “two-track” labor market, consisting of “professionalized” and “democratized” roles.

For “professionalized” roles, AI acts as a force multiplier, allowing the expert to accomplish more than they ordinarily would. These roles include radiologists and recruiters. On the other hand, in “democratized” roles, AI evens the playing field, making a role easier for non-experts to perform. These roles include IT service managers or medical secretaries.

Overall, the study finds “professionalized” roles are seeing greater wage and headcount growth, with twice the amount of available jobs and 42% faster salary growth compared to “democratized” roles.

AI skills such as prompt engineering are also becoming increasingly more desirable. Jobs that require specific AI skills are growing roughly eight times as fast as the overall jobs market, and the average wage premium for AI skills has risen to 62%, up from 57% the previous year.

Increase in Human Skills Required for Roles

Although leaders are no longer warning about an AI jobs apocalypse, the technology is changing and reshaping the skills needed to thrive in the workplace. PwC found human skills, such as judgement, creativity, and leadership are becoming more desirable.

This is particularly evident within entry-level hiring. Based on 2.4 million entry-level jobs in the US, those exposed mostly to AI are seven times more likely to require traditionally senior-level human skills such as leadership or face-to-face interactions. Job openings for these roles have grown 35% since 2019, while other entry level roles have shrunk by 10%.

“The companies seeing the greatest returns on AI are using it to amplify human expertise, accelerate innovation and create entirely new sources of value. As a result, they are pulling further ahead on productivity and growth than companies that focus primarily on automation.” – Joe Atkinson, Global Chief AI Officer at PwC

These findings suggest human skills and retaining that “human touch” remain critical in the age of AI, and that graduates entering the workforce for the first time will be tested on different grounds than they have been previously.

Major AI leaders have been very vocal in the past year about the impact AI will have on the job market. However, many are now reversing their original predictions, with Microsoft’s AI chief Mustafa Suleyman being the latest to course-correct.

This is coming at a time when AI could be causing more aches and pains than expected. Studies are mixed in how much AI has managed to increase productivity and return on investment so far, and AI blunders remain a big liability for businesses.

However, AI is still poised to sweep the business world and transform day-to-day work for employees. Going forward, businesses should prioritize the AI tools and workflows that directly address bottlenecks and compliment how they run.

AI Leaders Walk Away from Job Loss Narrative

Mustafa Suleyman, Microsoft’s AI chief is the latest AI leader to backtrack on the ‘AI jobs apocalypse’ narrative, with OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei both having done the same last month.

Suleyman said in February that the tasks involved with white-collar work, such as those done by accountants and lawyers, “will be fully automated by AI within the next 12 to 18 months.” The comment is similar to one made by Anthropic’s Amodei, who predicted AI has the potential to wipe out all white-collar jobs.

In a podcast released on Monday, Suleyman clarified his original statement, and said he expected AI to help white-collar workers complete tasks, rather than replace their roles altogether.

Is AI Performing as Expected for Businesses?

During a virtual interview at a Commonwealth Bank of Australia in May, OpenAI’s Altman said:

“I thought there would have been more impact on entry-level white-collar jobs being eliminated by now than has actually happened.” – Sam Altman, CEO of OpenAI

Even though there is some evidence to suggest entry-level work has been impacted by AI, Altman said his “intuitions were just off” in regard to the severity of the situation.

Overall, reports are conflicted about how much of a positive impact AI has been able to have on businesses so far. Some studies suggest an increase in productivity, with workers potentially saving as much as a full day of work when using the technology. However, business leaders have also been open about AI mistakes causing damage to their operations.

Businesses Will See Jobs Transform With AI

Businesses are still preparing for a future where AI completely transforms work. Leaders are preparing for reduced workforces and the chance to operate at a higher level of autonomy than before.

The technology isn’t yet perfect, but that hasn’t stopped the ongoing narrative about its potential. Entrepreneur Mark Cuban is noted in particular for calling AI stupid, but then suggesting businesses that don’t adopt it will fail.

Moving forward, businesses should prioritize the AI that works for them. Work closely with your team and your employees to develop a solution that fits with your operations, whether that’s by maintaining your company’s website with a vibe coding tool, or integrating agents to organize sprawling inboxes.

Anthropic has officially launched a new “safe” version of its flagship Mythos AI model. Named “Fable 5,” it is the first Mythos model to be released to the general public after its announcement in April 2026 prompted cybersecurity concerns.

Reportedly, Fable 5 has been built with guardrails to restrict its usage in “high-risk domains,” including cybersecurity. The company also plans to release an unrestricted version, known as “Claude Mythos 5,” to existing partner organizations.

Up until this point, access to Mythos has largely been restricted, with a select number of organizations gaining access to its highly advanced capabilities. The platform has received praise for its unrivaled ability to identify security vulnerabilities in critical infrastructure. It has the potential to help businesses push back against mounting data breaches, which have become rampant in recent years.

Anthropic Finally Releases Mythos to General Public

Anthropic has officially launched a restricted version of its Mythos model to the general public. Fable 5 is the first platform in the Mythos AI class to be made available to the public, despite its announcement back in April 2026.

At the time, the company revealed that it would be withholding access to the general public due to concerns that it might be used to commit widespread security breaches.

Instead, Anthropic shared Mythos with leading cybersecurity companies including CrowdStrike and Palo Alto Networks, part of an initiative that it dubbed Project Glasswing. The companies were full of praise for Mythos’s efficacy in detecting cybersecurity vulnerabilities.

Anthropic Restricts Access to Mythos ‘Sensitive Areas’

For the time being, the general public will not be able to access a full version of Fable 5. Rather, the existing version has restrictions on its usage in “sensitive areas,” especially its ability to identify cybersecurity vulnerabilities.

“The uplift from Mythos-level capabilities is valuable to many adversaries — for instance, those who could financially gain from cyberattacks — and we therefore expect them to be motivated to try to circumvent our safety measures.” – Anthropic spokesperson

Anthropic is also releasing a full version, known as Claude Mythos 5, to existing Project Glasswing partners. That list has been expanded to about 200 institutions across 15 different countries. The company is also working to deploy Claude Mythos 5 with the US government.

Mythos Could Overhaul Cybersecurity Infrastructure

Undeniably, Mythos has massive potential to help businesses everywhere keep up with a spiraling wave of cyberattacks. In recent years, data breaches have become rampant, with healthcare companies among those that are particularly vulnerable to exploitation.

The repercussions of a data breach can be catastrophic, with affected companies suffering financial penalties and reputational damage. In a lot of cases, firms are simply unable to recover from the dire impacts.

Mythos is uniquely-placed to help businesses identify critical security flaws that might otherwise have been exploited by bad actors. In an era in which companies are struggling to hire cybersecurity talent due to a massive skills gap, it could be particularly valuable in helping to turn the tide on cybercrime.

More than half of companies (53%) dealt with digital impersonations of their employees or executives within the past 12 months, a new report has revealed.

A large amount of the incidents were powered by AI tools. However, companies don’t seem to be prepared to deal with threats against the humans working at their operations, with 43% saying that they don’t run any person-of-interest threat profiles whatsoever.

The new data is yet another indication that, years after the technology turned into the tech world’s hottest buzzword, companies across all industries are still largely struggling to maintain cybersecurity protocols surrounding AI tools.

84% of Companies Have ‘Material Digital Risk Incidents’

The news comes from digital risk platform Outtake, which recently debuted the 2026 State of Digital Risk Report in partnership with Cybersecurity Insiders.

That report draws on a survey from early 2026 that logged responses from 1,138 decision-makers in charge of cybersecurity and fraud across a wide range of industries and sizes.

The biggest takeaway? Companies are not confident in their ability to address digital security issues.

Specifically, an enormous 84% of all organizations surveyed have experienced “material digital risk incidents” within the past 12 months.

At the same time, only 7% of these companies describe their digital security program as “leading,” and 21% don’t have a specific role for handling digital risk.

Impersonations Threaten Half of Today’s Businesses

The lack of preparation is a concern for a specific cyber attack that’s been making headlines: Impersonation, whether of employees or executives. This might be accomplished with AI-powered video or audio, or it might be an old-fashioned phishing email, but the end goal is to gain access to sensitive company data.

“Executive or employee impersonation hit 53% of organizations this year. Yet protection remains narrow: 77% limit workforce coverage to executives, a few high-risk roles, or reactive case-by-case response, and 43% run no person-of-interest threat profiling at all.” – the report

Companies are used to business-wide security, but aren’t prepared to create threat profiles for specific employees (43% have no capacity for this).

How AI Tools Create Impersonations

According to the Outtake report, 47% of respondents have “already encountered confirmed or suspected synthetic-media impersonation of an executive or brand representative.” In other words, a lot of the impersonations are AI.

Again, businesses don’t seem well prepared to handle this threat: 96% of them “have no automated way to stop an AI agent manipulated through its external inputs.”

The latest data indicates that humans working at companies constitute the most exposed and least protected attack surfaces. With AI-amplified threats only set to increase over the short term, there’s no denying that impersonations are among the biggest problems across many industries in 2026.

Hackers used AI to gain access to tens of thousands of high profile Instagram accounts last month.

No, they didn’t use an advanced AI model to steal credentials or generative tools to concoct phishing schemes aimed at breaching the accounts. Instead, hackers tricked the Meta AI support chatbot to give them access to all these accounts with nothing more than a simple prompt.

This news adds to the long list of security problems with AI, with experts pointing to a wide range of possible risks that could create problems for the tech industry and beyond.

Meta Confirms High Profile Instagram Hacks

Meta confirmed that, in late May, hackers gained access to a plethora of high-profile Instagram accounts, including Barack Obama and Sephora.

The social media giant reportedly patched the vulnerability earlier this month, although the number of hacked accounts remains a mystery.

“This issue has been resolved, and we are securing impacted accounts.” – Meta statement

How Did Hackers Gain Access to Instagram Accounts?

According to Meta, hackers were able to gain access to these Instagram accounts through the Meta AI support chatbot.

More specifically, hackers used a series of simple prompts to convince the chatbot that they had lost their passwords and needed to gain access to their accounts.

While Meta has some location-based safeguards in place to verify the identity of account holders, the hackers used a simple VPN to trick the Meta AI chatbot into believing the requests were legitimate.

The Security Risks of AI

In 2026, AI isn’t just for providing email summaries and generating funny images. Businesses are using agentic AI to automate entire departments with the ultimate goal of improving the heck out of productivity.

In that regard, though, AI is a double-edged sword. Giving AI — like the Meta AI support chatbot — access to that much information opens you up to potential breaches if you haven’t shored up your security.

If a company as big as Meta can be hacked through such simple means as a few prompts to a support chatbot, you better believe that investing in AI is no more important than investing in security for it.

Around a quarter of IT leaders report AI has already made mistakes that have negatively impacted their business, according to a new report.

Despite potential and realized business disruptions, AI investment remains optimistic. A large majority of IT leaders and employees said their company should maintain or increase its AI spending.

However, concerns remain around overreliance on AI and misuse, particularly as many leaders and employees said their company isn’t currently doing enough to encourage responsible AI use.

AI Mistakes Have Already Impacted Customers, Clients, and Bottom Line

Around a quarter of IT leaders say AI has made mistakes that has impacted customers, clients, and their company’s bottom line, in a new report from GoTo and Workplace Intelligence.

Similarly, even amongst the leaders who haven’t reported any impactful AI mistakes, concern remains high. Nearly all IT leaders surveyed (91%) worry AI could make a mistake that negatively impacts their organization.

Findings are based on a global survey conducted between November 2025 and January 2026. The sample included 1,250 full-time knowledge workers and 1,250 IT decision-makers.

Investment and Belief in AI Remains Strong

Despite concerns, companies aren’t slowing down on AI investments. Nearly all IT leaders surveyed (98%) said their company uses AI, and 82% of employees surveyed report using AI tools at work. Overall, 9 in 10 employees and IT leaders said their company should maintain or increase its AI spending, suggesting opinions of the technology are positive.

This is most likely the result of the significant benefits leaders and employees are already experiencing. 88% of employees said AI has benefitted them at their job. The main benefits include increased productivity (84%), better customer support (76%), and improved motivation and engagement (74%).

AI’s potential could even be greater still. The report suggests businesses could gain up to $2.9 trillion from unlocking AI’s full untapped value, and employees are still spending 2.6 hours per day on tasks that could be handled by AI.

Employees Remain Wary of AI Overreliance and Misuse

For all the spending on AI and its benefits, there are still major concerns among employees and leaders. Half of the surveyed employees say they rely too much on AI in their day-to-day, and 30% feel they can’t function without AI altogether.

Similarly, 39% of employees say relying on AI is making them less intelligent, suggesting there is yet to be a balance between using our own human judgement and relying on AI to do the thinking for us.

“AI has become a standard part of work, but our research shows that adoption alone doesn’t guarantee better outcomes. Employees are gaining productivity benefits, yet many are struggling with overreliance, inconsistent guidance, and a lack of training that helps them use AI responsibly and effectively. The organizations that will get the most value from AI are those that invest just as much in people, policies, and skills development as they do in the technology itself.” Dan Schawbel, Managing Partner at Workplace Intelligence

AI tools are also increasingly misused by employees. 70% of those surveyed admitted to misuing AI for sensitive or high-stakes work tasks. Overall, less than half of IT leaders said their company has an AI policy, and 84% of employees say their company isn’t doing enough to encourage responsible AI use.

AI is now commonly involved in serious data breaches and leaks, and so business leaders should ensure employees follow best security practices and that any sensitive data is well-protected.

Dashlane has been having a tough few days: The password management platform has suspended some customer accounts in the wake of brute force attacks on its service.

The event may already be in the rearview mirror, with the attacks and suspensions beginning in the afternoon of May 30 and the accounts being unsuspended by the evening of the same day.

In 2026, cybersecurity incidents are coming fast and furious, with voice-based threats and AI tools behind some of the most noteworthy events.

What to Know About the Dashlane Incident

The scale of the brute-force attacks is still unclear. However, multiple customers have received Dashlane’s automated emails alerting them that their accounts have been suspended in response to a third party’s repeated attempts to register a new device to the account without using the correct tokens.

As covered by The Register, these emails offer instructions on how to contact customer support and restore access to each account after explaining the problem:

“Your account has been temporarily suspended for security reasons as someone has attempted to register a new device and didn’t enter the correct token after several tries” -the emails

Dashlane has since sent a message out to social media, holding that “there is no evidence of compromise of Dashlane’s systems.” The platform says it is still monitoring the problem.

Some Login Attempts Hailed From Korea or Russia

Users have reported unauthorized login attempt notifications that give locations from other countries, with Korea and Russia among the top results, The Register reports.

It’s unclear if any these fraudulent logins were sucessful — Dashlane hasn’t commented.

Other users have voiced suspicions that the account suspension emails themselves are a phishing attack. However, based on Dashlane’s public responses, this doesn’t appear to be the case.

Dashlane is among the top password managers on the market today, according to our research, but this is undoubtedly bad PR for the brand.

Cybersecurity Incidents Remain a Top Concern in 2026

Dashlane’s struggles aren’t unique. There’s always a steady stream of data breaches and customer information losses across many industries, with finance and healthcare frequently among the biggest targets.

In 2026, however, top issues tend to center around AI tools – both due to the increase in hackers using these tools to more rapidly identify cracks in their target’s internet security and due to the increasing number of cracks caused by poor implementation of the cutting-edge tech.

Just a little over a month ago, for instance, we covered a car rental software company that saw its entire database wiped within seconds by its own agentic AI.

Outside of AI, one additional problem area for today’s cybersecurity experts are third-party integrations, with a recent report finding that one in four breaches used an exploit of a third-party vulnerability, as opposed to an internal attack.

We’d still recommend finding a password manager to add a layer of security — Dashlane’s automatic account suspensions are certainly annoying, but they’re better than a successful data breach.

99% of businesses are expecting AI to cause some headcount reduction in the next two years, according to a new report.

Likewise, business leaders believe redesigning work to fit around AI tools and automation is the best way to see a return-on-investment, suggesting significant structural changes are in the works for many teams.

Despite layoffs being an expected consequence of the AI boom, humans will continue to play a significant role in integrating the technology. Plus, many HR teams revealed a commitment to improving employee benefits and sourcing top talent.

Executives: AI Will Lead to Headcount Reduction

Mercer’s latest Global Talent Trends report found 99% of CEOs are preparing to cut staff due to advances in AI and automation.

Similarly, 98% of executives are planning organization design changes in the next two years, suggesting a massive shift in general business operations because of AI.

Mercer’s latest survey consisted of 9,000+ employees, 800+ C-suite members, 1,600+ HR leaders, and 100 investors.

Redesigning Work Around AI

Respondents of the survey also suggested that redesigning work in order to incorporate AI tools will be crucial for long-term success. Overall, 63% of executives believe a redesign initiative will drive the greatest return.

While only 32% of executives surveyed believe their current workforce is in a strong position to be combined with current automation technologies, employees remain eager to use these tools, and see them as a key part of their individual development. In particular, 35% of respondents said they’d consider quitting their role if they felt they didn’t have enough access to tools or training.

As well as redesigning work, the survey identified upskilling managers as another priority for business leaders. This will provide the skills needed to lead hybrid workforces consisting of both humans and AI agents.

Humans Remain Crucial

AI is already changing job roles, and layoffs remain unavoidable in some areas. Mercer’s data pinpoints those below the age of 27 as the most vulnerable, suggesting entry and mid-level work is most at risk.

However, recent data from Tech.co found businesses face a rework tax when using AI tools. Overall, SMBs spend over a quarter of an hour of AI use reviewing, editing, and fact-checking output. Those that spent the most amount of time reworking were also more likely to see higher productivity gains, suggesting humans still have a key role to play.

Recruiting and retaining top talent is a significant priority for business leaders in the next few years. 78% of executives say their organization is well-placed to compete for top talent, and 59% of HR leaders say difficulty attracting talent with vital digital skills is the top workforce challenge facing businesses in 2026.

Amazon is sharing the wealth when it comes to AI, now offering access to the new Agentic Shopping Assistant on AWS for retailers.

With the new tool, ecommerce businesses will be able to create custom AI chatbots that are familiar with the specific catalog, customer base, and shopping environment of your operation.

Agentic tools are becoming more and more common for businesses in 2026, and this tool will make it even easier for retailers to get provide a personalized experience to customers.

What Is the Agentic Shopping Assistant on AWS?

The Agentic Shopping Assistant on AWS is a customizable AI chatbot that can be deployed to different and specific retail businesses to help customers find specific products.

According to the AWS Marketplace overview, the tool is capable of providing “personalized product recommendations displayed as rich visual product cards—complete with images, pricing, and quick-reply options.”

The Agentic Shopping Assistant is built on Alexa for Shopping, which Amazon recently announced to help their own customers find products.

When Can Businesses Use the Agentic Shopping Assistant?

As Amazon points out in its announcement post, retailers “can’t afford to wait to build conversational shopping capabilities,” so how do businesses get a hold of this technology?

For starters, you have to be an Amazon Web Service (AWS) user. If you are, you can request a private offer on the AWS Marketplace or you can talk to your AWS account manager to get going with the tool.

Should Businesses Use the Agentic Shopping Assistant?

Increasingly, retail customers are interested in a tailored experience. In fact, according to one study, 83% of retail customers expect a personalized shopping journey.

That’s why the Agentic Shopping Assistant could be a big boon for retailers. It allows them to not only take advantage of modern AI technology, but also customize the experience for their specific brand.