OpenAI is launching the OpenAI Deployment Company in a push to help enterprises embed AI into their operations. The new company is designed to help organizations build and deploy their own systems to support “their most important work,” according to a statement.

Reportedly, the OpenAI Deployment Company will enable OpenAI to to embed specialized engineers into organizations that are seeking to solve complex issues. Known as Forward Deployed Engineers (FDEs), they will work with business leaders and frontline teams to identify where AI can be deployed.

ChatGPT, OpenAI’s flagship product, is still the de facto AI model of choice for many businesses, but Claude is one of the fastest growing and a favorite of developers.

OpenAI In Push to Help Enterprises Embed AI

OpenAI has announced the launch of a new company, OpenAI Deployment Company, as it gears up to help enterprises embed AI into their systems.

Reportedly, the new initiative will allow OpenAI to deploy more specialized engineers, known as Forward Deployed Engineers (FDEs).

Specializing in “frontier AI deployment,” these engineers enable businesses to solve “complex problems in demanding environments,” according to a statement. They will work in close collaboration with both frontline teams and business leaders to identify areas where AI could be deployed to solve problems.

The new company has launched with $4 billion of investment at a $10 billion pre-money valuation, with backers including Goldman Sach, Bain & Co., Capgemini, and McKinsey & Co.

OpenAI Acquires AI Consulting Firm

OpenAI has also acquired Tomoro, an applied AI consulting and engineering firm. As part of the move, 150 FDEs will be made available to OpenAI Deployment Company, which in turn will help more organizations to embed AI into their workflows.

Ultimately, OpenAI has stated that its decision to launch a new deployment vehicle is in the service of “building for where frontier AI is headed.” Not only will it help new businesses to embed AI into their systems, but it will also operate as an extension of OpenAI, continuing to research, develop, and launch the latest AI technology.

“AI is becoming capable of doing increasingly meaningful work inside organizations. The challenge now is helping companies integrate these systems into the infrastructure and workflows that power their businesses. DeployCo is designed to help organizations bridge that gap and turn AI capability into real operational impact.” – Denise Dresser, OpenAI Chief Revenue Officer

ChatGPT Still Winning the Chatbot Race

Since launching in November 2022, OpenAI’s ChatGPT has enjoyed first-mover advantage and, subsequently, a dominant market share among users and businesses alike. And with adoption increasing at a scarcely believable pace, this trend is unlikely to abate.

More broadly, the AI race continues to unfold at breakneck speed. The recent explosion of different vibe coding platforms has opened up another avenue for users to explore, with tools such as Base44 and Lovable allowing non-technical users to code their own websites and apps through the use of AI.

All of this deployment comes at a cost — more and more, businesses around the world are laying off staff in favor of automation. While many people will try to deny the link between the two phenomena, it’s becoming increasingly difficult to ignore.

Google has just reported the first (known) example of hackers using AI to uncover and exploit a zero-day security flaw.

It’s a sign that the AI threat many experts predicted is officially here for the cybersecurity industry: AI tools can help bad actors discover new pathways towards their goals.

Here are the details to know.

How AI Developed a Zero-Day Exploit

The news comes from a freshly released report by the Google Threat Intelligence Group (GTIG) that follows up on the group’s February 2026 report covering AI-related threat activity.

Now, they say, they have for the very first time “identified a threat actor using a zero-day exploit that we believe was developed with AI.” Specifically, it was within a Python script that let the user bypass two-factor authentication on “a popular open-source, web-based system administration tool.”

The report says that it likely stopped this particular threat:

“The criminal threat actor planned to use it in a mass exploitation event but our proactive counter discovery may have prevented its use. Threat actors associated with the People’s Republic of China (PRC) and the Democratic People’s Republic of Korea (DPRK) have also demonstrated significant interest in capitalizing on AI for vulnerability discovery.” – GTIG

For hackers, the report concludes, AI tools represent “expert-level force multipliers for vulnerability research and exploit development.”

How Google Identified the Use of AI

You might be wondering what makes Google so confident that the hackers relied on AI specifically to uncover the exploit itself – after all, the code didn’t appear to have included the word “delve.”

However, it did include other tell-tale signs, Google says.

“[T]he script contains an abundance of educational docstrings, including a hallucinated CVSS score, and uses a structured, textbook Pythonic format highly characteristic of LLMs training data (e.g., detailed help menus and the clean _C ANSI color class).” -GTIG

The threat was stopped, as GTIG worked with the impacted vendor to disclose the problem.

And don’t worry, Google fans: Their team says that they “we do not believe Gemini was used” for this particular process.

The Age of AI-Powered Cybercrime

AI agents and other LLM-powered tools have been making waves across all areas of business security for years now.

That’s not all good news. Recently, Anthropic’s Cursor wiped a company’s entire database within just a few seconds — before then issuing an apology.

That’s not stopping the rollout of even more AI. At least, not according to a new Gartner prediction that claims the average global Fortune 500 company will have over 150,000 AI agents by 2028.

For cybersecurity professionals, increasing AI use means that criminals and security experts alike have a new set of tools to help each group attempt to one-up the other.

We can only hope the bad actors are at least as likely to accidentally delete their entire database as all the law-abiding professionals.

Anthropic has revealed that its flagship Claude chatbot will soon be able to “dream” in a process mirroring what the human brain does when we sleep. Specifically, it will go over recent events and identify things that are worth committing to “memory,” which can then inform future tasks and interactions.

Unlike compaction — the process that all AI agents undergo in which conversations are regularly audited for irrelevant information — dreaming periodically recurs across multiple agents. This will identify complex patterns that a single agent couldn’t identify alone.

Dreaming is currently limited to Claude Managed Agents, which “provides the harness and infrastructure for running Claude as an autonomous agent,” according to the company. Reportedly, the process will be invaluable for long-running, complex projects that involve multiple AI agents.

Claude Managed Agents Can Soon Dream

Anthropic has revealed that Claude will soon be able to dream. In a process closely mirroring what the human brain does when we sleep, the chatbot will soon revisit recent interactions and identify specific things that are worth storing in its “memory.”

The feature is currently reserved for Claude Managed Agents, which are pre-built agents that operate in managed infrastructure. They’re engineered for long-running and complex tasks that require a high degree of asynchronous work.

Anthropic revealed the news at its latest Code with Claude developers conference. It is still in research preview, but interested developers can request access.

Dreaming Will be Invaluable for Complex Processes

Regular chatbots already undergo a process known as compaction, in which lengthy conversations with users are regularly analyzed, and the models remove information that is irrelevant to the context window. However, Anthropic claims that dreaming differs in a few key ways.

To begin with, while compaction is limited to a single conversation with a single agent, dreaming allows conversations across multiple agents to be analyzed and stored. In theory, this will enable the agents to identify complex patterns. It could be a game changer for developers working on multi-step projects that require a high level of asynchronous work.

“Dreaming surfaces patterns that a single agent can’t see on its own, including recurring mistakes, workflows that agents converge on, and preferences shared across a team. It also restructures memory so it stays high-signal as it evolves.” – Anthropic spokesperson

Claude the Go-To Platform for Developers

From a developer standpoint, Claude continues to go from strength to strength. The platform is already considered the de facto solution for coders everywhere, with the company announcing recently it is rolling out AI agents for financial services.

AI is already deeply embedded in software development, with 90% of developers now using the technology at work, and this trend is showing no signs of abating, with AI shown to accelerate coding. The rollout of various vibe coding platforms has taken this even further.

This comes in spite of various warnings from leading figures in the tech space that mass unemployment due to AI is likely.

Anthropic is rolling out even more AI business tools, with the company announcing new agents that are specifically aimed at improving financial services.

These ready-to-run templates will be able to perform a wide range of tasks, from preparing meetings and building pitches to reviewing valuations and auditing statements.

With the release of these ten AI agents, Anthropic continues to strengthen its enterprise offerings, which have already become the go-to option for businesses in 2026.

Anthropic Launches New AI Agents for Financial Services

Announced in a company blog post, Anthropic unveiled ten new, ready-to-launch AI agent templates that are specifically designed to handle one of the most important aspects of any business.

“Finance is a great blueprint for the rest of knowledge work.” – Nicholas Lin, Anthropic’s head of product for financial services to Bloomberg

There are ten new financial service templates, which can perform and automate a wide range of tasks, including meeting preparation, statement auditing, earnings review, and market research.

How Do Anthropic’s Financial Service Agents Work?

There are two distinct ways in which you can utilize these new AI agents from Anthropic to make financial tasks a lot easier.

For starters, each of the ten agent templates are available as a plug-in for Claude Cowork and Claude Code, with these tools running alongside the analyst to provide more specific assistance depending on the agent.

Additionally, you can use any of these agent templates as a cookbook for Claude Managed Agents, which runs autonomously alongside the Claude platform to streamline your financial operations.

Claude for Business Is Booming

We’ll just come right out and say it: Claude is becoming an absolute powerhouse for businesses and the enterprise industry is taking notice.

The launch of Claude Cowork and Claude Code in January has spurred some serious interest in the AI model, with Anthropic now accounting for 73.3% of new AI spending, compared with only 26.7% from OpenAI.

With these new financial service agent templates, Claude could be positioning itself as the best option for businesses for the foreseeable future.

Check out our Claude for Business guide to learn more

Approximately 80% of the businesses that have piloted or deployed autonomous business capabilities have reported workforce reductions, a new survey reveals.

However, workforce reductions aren’t causing direct return on investment, and it’s predicted over time that autonomous businesses will generate more opportunities for the job market.

Several other surveys suggest human input remains a critical factor in determining how successful businesses are with their AI investments.

Business Using Automation Are Reducing Workforces

According to a new survey conducted by Gartner, around 80% of organizations piloting or deploying autonomous business capabilities have also reported reductions in their workforces.

The adoption of technologies such as AI agents, intelligent automation, digital twins, and tokenized assets will turn businesses from simply augmenting certain processes into truly autonomous organizations, says Gartner.

Findings are based on a survey of 350 global business executives, within organizations reporting enterprisewide revenue of at least $1 billion or equivalent. They had been piloting or had already deployed at least AI agents, intelligent automation, or autonomous technologies in the third quarter of 2025.

Workforce Reductions Don’t Translate to ROI

Despite the high number of organizations making reductions, this didn’t directly translate to return on investment (ROI). This suggests that replacing human workers with autonomous systems isn’t the best move for businesses wanting to see maximum ROI.

“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.” – Helen Poitevin, distinguished VP analyst at Gartner

Despite the findings above, Gartner predicts autonomous businesses will “create more work for humans, not less,” says Poitevin. This is because they will create new forms of work that AI won’t necessarily be able to absorb.

Likewise, Gartner forecasts organizations will be spending $168.9 billion more on AI agent software from 2026 to 2027. In 2025, the amount of overall spending was only $86.4 billion.

Human Talent Remains Essential for Autonomous Businesses

Ultimately, these findings lend themselves to the theory that human talent remains an essential building block for businesses looking to form their AI castles.

“Lasting structural factors such as demographic decline and high-stakes, trust-dependent consumer moments will ensure human talent remains central to running, governing and scaling autonomous businesses.” – Poitevin

Several studies corroborate the fact that some processes need to remain human-led in order to be successful. For example, a study on email marketing preferences among consumers found that communications that ‘felt human’ were preferred.

Likewise, recent data from Tech.co found that businesses are spending 26% of an hour of AI use reworking output. While this suggests AI slows down productivity, we found that businesses spending more time in the rework phase were more likely to see AI-related productivity gains. This suggests that keeping humans integrated into autonomous systems can be beneficial to overall productivity.

Education technology firm Instructure has confirmed a data breach due to a cyberattack, citing “disruption to tools relying on API keys” as the source of the breach.

The US-based company is most well-known for Canvas, a widely used learning management platform that tracks online coursework and assignments.

The ShinyHunters extortion gang has since claimed responsibility, listing the company on its data leak site while claiming to have data from “nearly 9,000 schools worldwide.”

How Did Instructure Handle It?

Instructure first acknowledged the incident on April 30th, saying that their “team is actively investigating and has taken precautionary steps to help maintain service stability while we work to restore full functionality.”

The next day, they issued a more detailed statement, including a list of steps they’ve taken to address the issue:

• Revoked privileged credentials and access tokens associated with affected systems.
• Deployed patches to enhance system security.
• Out of an abundance of caution, rotated certain keys, even though there is no evidence they were misused.
• Implemented increased monitoring across all platforms.

What Data Was Leaked?

The company didn’t list the number of affected institutions, but it did name the type of data that they believe was breached, which includes:

• Names
• Email addresses
• Student ID numbers
• Messages among users

Not included? According to the company, it has “found no evidence that passwords, dates of birth, government identifiers, or financial information were involved. If that changes, we will notify any impacted institutions.”

What Does ShinyHunters Have to Say?

The extortion group called ShinyHunters claimed credit for the breach on May 3rd, adding Instructure to its Tor-based data leak site, SecurityWeek reports.

ShinyHunters has more specific claims surrounding the full scale of the breach. The group states that:

• 3.65 terabytes of data has been stolen.
• 275 million students, teachers, and other staff members impacted.
• “Several billions of private messages” stolen.
• Nearly 9,000 education institutions worldwide impacted.
• Instructure’s Salesforce instance is compromised as well.

It’s another example of a major data breach emerging from software vulnerabilities related to third-party apps — we’ve covered cloud hosting platform Vercel’s struggles with a similar issue several weeks ago, and studies show that one in four modern breaches exploit a third-party vulnerability.

You can now build an app with an app, because vibe coding platform Lovable has officially added its platform to the Google Play Store and the Apple App Store.

The app will have limited functionality compared to the desktop version of Lovable, but users will be able to queue prompts to generate your app and get notifications about the progress of your creation.

Lovable did have to jump through some hoops to make this happen, in the wake of Apple cracking down on vibe coding tools for iOS users.

Lovable Launches Mobile Vibe Coding App

Announced in a company blog post, Lovable is officially launching a mobile app that will allow users to generate apps on the go.

“Whether you’re a founder prototyping a new concept, a designer testing a layout, or someone who’s never written a line of code — you can turn ideas into real, working web apps straight from your phone.” – Lovable announcement

The app is currently available on both the Apple App Store and the Google Play store right now.

What Can the Lovable Mobile App Do?

Obviously, mobile apps are rarely as powerful as the actual platform, but in Lovable’s case, the app can perform many of the tasks that the desktop platform can. Here are some of the functionalities available:

• Queue prompts – You can input voice or text prompts into the Lovable app and queue them up for action while you’re on the go.
• Get notifications – The Lovable app will send you notifications to update you on the progress of your generated app.
• Seamlessly switch – Swapping between desktop, laptop, and smartphone is easy now, as the Lovable app syncs all your changes across devices.

Simply put, the Lovable app makes it easier for you to instantly act on your inspiration, no matter where you are.

App Launch Amidst Vibe Coding Crackdown

The Lovable app is, in fact, available on the Apple App Store, which is a big deal considering Apple has been cracking down on on vibe coding apps over the last few months.

Apple has noted that the removal of vibe coding apps and the blocking of updates on apps like Replit is because these apps can be edited after they have been published. Given Apple’s stern protection of its app store through its review process, this kind of circumventing the rules could not stand.

Lovable played by the rules, though, generating web-based applications on its platform rather than downloading new executable code, which is why it’s now available as a mobile app on both app stores.

An AI agent mistakenly deleted a company’s entire customer database — and then issued an apology afterwards. The culpable AI agent was handed a routine task, to which it responded by wiping the company’s data.

The company, PocketOS, which makes software for car rental businesses, experienced a 30-hour outage when the AI tool erased its database. The tool in question was Cursor, a popular vibe coding agent powered by Athropic’s Claude 4.6 model.

The incident is not the first high-profile error that has come about because of AI. With development and adoption continuing at breakneck pace, companies face growing pressure to introduce safeguards to ensure gaffes such as this are few and far between.

AI Agent Wipes Entire Company Database, Apologizes

An AI agent designed to streamline basic coding tasks wiped out a company’s entire database — and then issued an apology. The company tasked the AI agent with a basic action, and it responded by erasing its database.

According to PocketOS founder, Jer Crane, when asked to explain its action, the agent “produced a written confession enumerating the specific safety rules it had violated.” Reportedly, the agent had ignored a key safeguard preventing “destructive” or “irreversible” actions without explicit permission.

The incident, which took just 9 seconds, led to a major 30-hour outage over the weekend. Luckily, the company was able to recover its data two days later.

Cursor Agent Responsible for Database Wipe

The responsible agent was Cursor, a vibe coding platform that is powered by Anthropic’s Claude Opus 4.6 model. Cursor is favored by skilled developers as a tool for accelerating the coding process.

According to Crane, the AI stated that: “Deleting a database volume is the most destructive, irreversible action possible — far worse than a force push — and you never asked me to delete anything. I decided to do it on my own to ‘fix’ the credential mismatch, when I should have asked you first or found a non-destructive solution.”

Due to the outage, companies using PocketOS temporarily lost access to customer records and bookings. Crane stated: “Reservations made in the last three months are gone. New customer signups, gone.”

A Note of Caution to Businesses Everywhere

The PocketOS data incident should sound alarm bells throughout the tech sector, which has moved swiftly to embed AI into its ways of working in the last few years.

As observed by Crane: “This isn’t a story about one bad agent or one bad API. It’s about an entire industry building AI-agent integrations intro production infrastructure faster than it’s building the safety architecture to make those integrations safe.”

With innovation and adoption continuing, businesses face mounting pressure to introduce the requisite safeguards to ensure events like this are not a common occurrence. After all, downtime and compromised cybersecurity can have a catastrophic impact on a business’s reputation, not to mention its finances.

In a new report, Gartner has predicted that the number of AI agents within the average global Fortune 100 enterprise will exceed 150,000 by 2028, suggesting a huge increase from 2025.

To help businesses combat the risk of AI sprawl, Gartner has identified six steps businesses can take to manage and govern an increasing number of agents effectively.

While businesses should actively encourage employees across teams to experiment with AI agents, having the proper governance procedures in place is critical to ensuring safety.

Number of AI Agents Expected to Skyrocket by 2028

A new report from Gartner sees the average global Fortune 500 company using over 150,000 agents by 2028, which is up significantly from 15 in 2025.

This will, Gartner predicts, generate agent sprawl, defined as an uncontrolled accumulation of AI agents built by different teams without centralized governance or consistent oversight. AI sprawl can lead to misinformation, oversharing, and data loss.

Similarly, an increase in the number of AI agents can lead to IT complexity and management challenges.

Gartner Identifies Six Steps to Avoid Agent Sprawl

To help business leaders reduce the risk of agent sprawl, Gartner has identified six steps that will help businesses build a strong foundation for onboarding new agents.

• Establish agent governance and policies – Set clear rules for when and how agents are built within your company. Likewise, outline who can build agents, who can share them, and what connectors are permitted.
• Build a centralized agent inventory – Consider using different AI management tools to categorize agents across applications, from both sanctioned tools and shadow AI solutions. With an agent inventory, businesses can build adaptive controls to enforce the right policies based on the level of risk each agent poses to their business.
• Define agent identity, permissions, and life cycle – Manage each agent’s identity, permission model, and access controls. Ensure you’re continually reviewing and retiring redundant agents to prevent an uncontrolled sprawl.
• Create AI information governance – Control what information the AI tool or agent has access to. Implement a continuous process to keep all data current, manage data permissions (to prevent oversharing), and archive obsolete data.
• Monitor and remediate agent behavior – Establish ongoing visibility into how agents are being used. Put processes in place to ensure policy compliance, detect anomalous behavior, and correct agents that act beyond or in breach of their established guidelines.
• Encourage responsible AI usage amongst employees – Support teams and employees with AI training programs and develop a positive community to encourage adoption and best practices across your business.
  

The Value of Structured AI Governance

Gartner found that only 13% of organizations think they currently have the right AI agent governance in place. Additionally, similar studies have warned against the potential of AI-driven threats to today’s businesses.

According to Max Goss, senior director analyst at Gartner, businesses should avoid “blocking or restricting” access to certain AI agents and tools, as employees will simply go around restrictions. This will ultimately present more risks.

Instead, Goss says: “Organizations need to find a balance where they can govern agents and manage sprawl, but also safely empower employees to innovate with these tools.”

Over a quarter of CEOs surveyed in a recent study expect their organizations to act primarily without human intervention by 2028, due to advances being made in automation and AI.

The study suggests that CEOs predict to move to a more autonomous business structure as AI and automation capabilities grow, particularly in areas of human decision-making.

On the other hand, the CEOs surveyed don’t anticipate any changes to their customer base as a result of the rise of AI. Instead, businesses are using the tool to retain and engage customer relationships.

CEOs: AI Will Enable Decision-Making Without Humans

According to a new study from Gartner, 27% of CEOs expect their organizations to operate mostly without human intervention in 2028, because of AI developments and growing automation capabilities.

David Furlonger, distinguished VP analyst at Gartner, suggests that the results show, “CEOs are realizing that AI is not simply another layer of automation. It is a catalyst for rebuilding the enterprise itself.”

The findings are based on a survey of 469 CEOs and other senior business executives around the world. The survey ended in the fourth quarter of 2025.

AI Will Turn Digital Businesses into Autonomous Ones

The value AI could bring to enterprises hasn’t passed business leaders by. Overall, 80% of the CEOs surveyed expect AI to enact a high to medium degree of change to their companies’ operational capabilities. Gartner predicts this will cause businesses to move from the structure of a digital business to an autonomous one.

“Autonomous business is a strategy where self-learning software agents and machine customers make decisions, take action and create new types of value for organizations… While digital business changes what the organization does, autonomous business changes how the organization does it.” – Don Scheibenreif, distinguished VP analyst at Gartner.

Further findings make it apparent that CEOs and business leaders only expect their automation capabilities to grow. 54% of CEOs reported that their automation was limited to specific tasks, and only 13% expected it to remain at this level, by the end of 2028.

Similarly, 32% of CEOs expect their organizations to integrate self-learning and adaptable AI tools to assist with human decision-making. This certainly signals that the scope for automation, in the eyes of today’s CEOs, remains wide.

Customer Bases Will Remain Unchanged Due to AI

Recent studies have suggested that consumers want a human touch, even in the wake of AI developments. Customers still value the relationships they create with companies, and only 17% of the CEOs surveyed expect significant changes to their customer base because of AI.

In fact, the study found CEOs and business leaders are using AI to deepen their relationships with customers and increasingly, machine customers.  Through 2026, Gartner predicts the number of large companies with a dedicated sales channel to access fast-growing machine customer markets will double from 2024.

However, while AI may assist with processes like decision-making and customer engagement, humans remain a necessary part of the process for training and developing the technology. What CEOs and business leaders should focus on is solidifying their existing people, assets, and financial structures to withstand the incoming changes.

Employees are being left behind when it comes to the AI revolution, with only 15% of workers stating that they have access to AI training tools at their jobs.

This is likely exacerbating the fear of AI in the workplace, with 10% of workers believing that they will be unemployed without AI training, according to the study.

These stats clearly illustrate the need for AI training in the workplace, particularly if your business is investing in technology for future success.

AI Training and Job Security

According to a study from Federal Reserve Bank of New York, a mere 15.9% of US workers reported that their employer currently offers AI training tools for employees at their business.

“Although most workers with access to AI tools recognize the productivity-enhancing impacts of AI, training in AI tools is not available to all of them.”

Even worse, workers want to learn how to use AI to improve these companies, with 38% of employed respondents noting that it is important to them to have training in how to use AI tools.

AI Users Are Higher Earners

The concern about access to AI tools is a very real one, particularly because those getting AI training and using AI tools are doing better at their jobs than those who aren’t.

The survey found that 66.3% of workers earning over $200,000 per year are using AI, while only 15.9% of workers earning under $50,000 per year are using AI.

Survey results also found that full-time workers (42.7%) are using AI more frequently than part-time workers (24.7%) and college-educated employees (58.7%) are using it more than non-college-educated employees (22.9%).

How to Provide AI Training for Employees

If you want to avoid being one of those businesses that isn’t providing AI training tools for your employees, you are in luck, because it’s easier than ever to get your staff educated on this new technology.

There are dozens if not hundreds of free AI training courses from education providers that can get you started. They focus on everything from prompt engineering to AI literacy, so your team can be highly functional with the tech.

It’s important to remember that, with AI, replacing humans isn’t an option just yet. The businesses that are succeeding the most are the ones using AI to augment and improve human processes, rather than ditching their workforce for the tech.

Cloud hosting platform Vercel has just revealed a big internal data breach.

The security incident stemmed from a breach in a third-party AI tool’s Google Workspace OAuth application, which threat actors used to access an employee’s Workspace account.

The company has confirmed that its popular open-source projects, including Next.js and Turbopack, remain secure.

What to Know About the Breach

The breach was due to an app made by Context AI, Vercel says, which one employee downloaded and connected to their corporate account.

In its statement about the incident, Vercel said that the credentials for “a limited subset of customers” were compromised. Anyone who hasn’t already been contacted was not in that group, they say.

Vercel has “deployed extensive protection measures and monitoring,” the company added.

A Hacker Claims to Be Selling Breached Data

There’s more to the story: A threat actor calling themself “ShinyHunters” has taken credit for breaching Vercel in a hacking forum, Bleeping Computer reports.

The hacker also claims to be selling the stolen data, which reportedly includes:

• Access keys
• Source code
• Database data
• Internal deployments
• API keys

The forum post further explains that, “the access I’m about to give you includes multiple employee accounts with access to several internal deployments, API keys (including some NPM tokens and some GitHub tokens).”

These claims haven’t been verified, however, so we don’t know for sure how much of what the hacker is saying is true.

Supply Chain Hacks on the Rise

The incident is an example of the growth of a type of hack that targets software developers that provide commonly used code. In this case, Vercel is behind the popular React framework Next.js.

Granted, this framework wasn’t impacted in this particular hack, but if successfully compromised, the software could open up a broad range of targets for hackers hoping to access data at scale. After all, the Vercel breach emerged through a breach at Context AI.

CEO Guillermo Rauch addressed the incident on his personal social media account, saying that “my advice to everyone is to follow the best practices of security response: secret rotation, monitoring access to your Vercel environments and linked services, and ensuring the proper use of the sensitive env variables feature.”

We’ve previously reported on a study that found one in four data breaches are due to exploits of third-party apps. Looks like we can now add one more example to the list.

According to a new survey from Tech.co, small to medium-sized businesses (SMBs) spend over a quarter of an hour (26%) of their AI use reworking output, including editing and fact-checking.

The businesses spending the most amount of time reworking, however, were more likely to see higher AI-related productivity gains.

These findings reveal the indispensable role humans are currently playing as part of AI systems in enterprises, and they remain critical as cybersecurity and quality control concerns mount.

SMBs Spend 16 Minutes Per Hour Reviewing AI Output

Tech.co’s latest survey shows that for every hour of AI use, SMBs reportedly spend an average of 16 minutes reviewing, editing, or fact-checking AI output.

This amounts to over a quarter (26%) of time within the hour. This suggests that there is a “rework tax” businesses are paying in order to reap the full benefits of AI systems within their enterprises.

Results are derived from a recent survey of 300 SMB owners and C-suite members.

5 to 25 Minutes of AI Rework Time Is Optimal for Productivity

Our research shows that businesses achieving the highest levels of productivity, saving 11+ hours or more per week, actually dedicated more time to reviewing AI output compared to the least productive groups.

59% of the group with the lowest rework time, less than 5 minutes, reported only 2 hours saved per week.

On the other hand, 20% of the group with a higher productivity saving of 20+ hours per week reportedly spend 25-35 minutes reworking AI output, and 13% spend 25-55 minutes. The highest productivity group were also the most likely to spend 55+ minutes on rework (13%).

Overall, the very existence of this “rework tax” shows AI is yet to be a complete, capable product on its own. In our survey, the businesses that understood this the most, by dedicating the necessary time to refine and adjust AI output, were more likely to see productivity benefits.

A Human in the Loop Is Still Necessary for Productivity Gains

If SMBs want to successfully adopt AI and reap its benefits, using human judgement to rework output is an operational necessity. In its current state, AI systems are unable to produce output that is completely correct and aligned all of the time, meaning that if businesses want to dodge quality control concerns, human reviewers are vital.

Likewise, cybersecurity within AI systems remain a critical concern for modern day SMBs. Alongside general cybersecurity best practice, keeping a human involved ensures systems behave and use data safely, via continuous testing and reviews.

However, businesses should also consider the potential effects of this “rework tax” on employees, who may experience fatigue or frustration if they are always correcting AI systems.

AI is finally starting to carry its weight, with new data from Tech.co discovering that 22% of small businesses owners are saving between six and ten hours per week by using AI.

That’s not all, though. A majority are seeing productivity gains, with 54% of small business owners stating that AI technology has improved their operation in some capacity since implementation.

As for what they’re using AI for, the survey found that writing tasks and research are quite common, while customer support remains one of the most productive industry for AI.

AI Is Improving Productivity

With 22% of businesses saving six to ten hours per week by using AI and 54% of business owners seeing productivity boosts, it’s safe to say that the technology is finally contributing in a serious way.

In fact, our data found that those spending heartily are seeing some of the biggest gains. Businesses spending between $1,001 and $2,500 on AI per month are the ones typically saving a full day of work, while businesses spending under $100 per month are saving less than 2 hours per month.

All in all, AI remains a classic get-what-you-pay-for situation, with free and entry-level solutions not providing much when it comes to enterprise-level productivity gains.

What Are Businesses Using AI For?

We also asked business owners what they’re actually using AI for, so that we can get a feel for how the technology can best be utilized for productivity gains.

Here are some of the top tasks businesses leaders are automating in 2026:

• Writing tasks – 29%
• Research – 26%
• Customer support – 19%
• Taking meeting notes – 16%
• Scheduling & calendar management – 16%
• Money & bookkeeping – 15%
• Staffing & team admin – 13%

Right now, it seems like businesses are only comfortable automating some of the more basic tasks, while more advanced, human-level requirements like money and staffing are still waiting in the wings for more capable AI tools.

What Tasks Are AI Best For?

While we did provide you with some information about which tasks are being automated the most, it’s also worth understanding what automated tasks are improving productivity the most.

Our data found that 80% of business leaders who automated customer support tasks with AI saw improvements in productivity, representing the largest percentage of business leaders doing so.

Essentially, when speed and efficiency are the most important means of accomplishing a task, as is the case with almost all customer support interactions, AI tools can make a big difference for your business.

A certain mindset is taking over the logistics industry, with recent data from Tech.co surveys finding that growth has become more of a priority than ever.

As for what kind of strategies they’re employing to do so, logistics businesses were primarily found to be adopting new technology and optimizing recruitment in order to facilitate this growth.

Why are businesses more focused on growth than ever before? Well, you’ll have to keep reading to find out!

Record-Breaking Rise in Logistics Businesses Prioritizing Growth

According to Tech.co’s Logistics Growth Stability Index, more logistics businesses are prioritizing growth than ever before.

The index rose to 1.3 in March 2026, marking the highest level since we started tracking in March 2025.

More notably, the rise comes after February 2026 saw the lowest prioritization of growth since we started tracking, coming in at only 0.8.

Growth Strategies Include New Tech & Recruitment

Prioritizing growth is one thing, but how are logistics businesses in 2026 actually trying to grow their companies right now?

According to our data, 19% of businesses are prioritizing new technology as a way to spur growth. That represents the most common strategy for businesses in our data, and it shows a 2% increase since February, when only 17% of businesses said the same.

The second biggest priority for growth in logistics businesses is staffing, with 18% of businesses stating that increasing retention and recruitment were the means by which they plan to grow in 2026.

Why Are Logistics Businesses So Focused on Growth?

There are three big reasons why we think businesses in the logistics industry are so focused on growth, specifically in March 2026.

• High freight demand – 83% of US logistics businesses had a moderate to high level of freight to haul in March, so operations need to grow to manage it.
• Regulatory concerns – New regulations that ban non-English speaking drivers could be fueling a need to hire more drivers and retain current staff.
• Volatile fuel costs – Fuel card adoption has risen 10% in the last month, which could point to rising fuel costs forcing businesses to find ways to earn more through growth.

The logistics industry is always in flux, but these changes are far more substantial than usual. If your operation is in need of some growth, be sure to check back with Tech.co for more logistics stats on how the industry is adapting.

Higher education facilities are increasingly implementing work-based learning in cybersecurity programs in an effort to overhaul the yawning cybersecurity skills gap, new research finds.

According to a study conducted by the University of South Florida, colleges and universities are determined to close the skills gap and have begun to incorporate hands-on learning into their existing cybersecurity programs. Practitioners argue that such an approach can create a more “scalable” and “career-ready” talent pipeline.

Cybersecurity is a growing concern among businesses around the world. The rapid adoption of AI has opened up new attack surfaces for cybercriminals, as well as giving rise to a new generation of sophisticated tools that are difficult to detect. The skills gap is commonly cited as one of the biggest obstacles in countering this evolving threat landscape.

Higher Education Firms on a Mission to Close Cybersecurity Skills Gap

Higher education facilities across the US are increasingly deploying work-based learning into their cybersecurity programs as part of an effort to address the industry skills gap, according to a new report.

With many businesses around the world struggling to cope with an escalating wave of cyberattacks, the study highlights the importance of hands-on learning to better prepare students for the future workforce.

“This is critical in an environment where technology is rapidly changing and the workforce has to adapt to those changes. Strengthening programs with the use of work-based learning can create a more scalable, inclusive, and career-ready cybersecurity talent pipeline.” – Oscar A. Aliaga, Associate Professor, University of South Florida

Hands-On Learning Proliferating Across Facilities of Different Types

To gather their findings, the researchers mapped out work-based learning (WBL) practices with a cross-sectional survey. Mainly, they were preoccupied with prevalence, regional distribution, and the characteristics of such an approach.

Ultimately, it was found that, while WBL opportunities span all levels of higher education credentials, they are most prevalent among certificates and associate degree programs. This suggests that educational practitioners are eager to embed hands-on cybersecurity learning at foundational levels.

Elsewhere, the research identified 266 WBL activities across a wide variety of types, including industry speaker series, capstone experiences, worksite visits, job shadowing, employee mentorships, and senior design projects.

Closing Skills Gap Key to Combating Mounting Security Threats

As we’ve covered elsewhere, global cybersecurity has reached a “tipping point,” with the rise of AI being a significant contributing factor. To make matters worse, businesses are understaffed and under-resourced where it matters most: their IT and security departments.

According to a recent report, 88% of professionals have experienced at least one “significant” cybersecurity consequence as a result of the skills gap. Concerningly, the data indicates that the skills gap is actually getting worse, with 95% of surveyed respondents reporting at least one skill need, up from 90% the year before.

And with 68% of employees believing that existing security measures are insufficient to deal with the rising threat level, the need for cybersecurity talent has never been greater. The pervasiveness of WBL practices across the nation poses hope for the future.

Tax season is a holiday for many hackers: If they’re after financial information, there’s no better time to wheedle it out from unsuspecting victims.

In 2026, there’s a new scam threat to consider, and that’s AI. Phishing attempts might use text-based LLMs or voice-based impersonations, but with AI powering them, bad actors can send a hundred times as many messages.

Here are the signs to look for, as well as the ways that experts like the Better Business Bureau (BBB) are recommending people respond in order to stay safe.

A ‘Record Number of Calls’ to the IRS

Many of the most recent tax-related AI scams are fear-based: They impersonate debt collectors or the IRS, claiming that their target needs to immediately pay a fictional debt.

Threats of stronger penalties or even jail time help to keep the target too scared to question whether the phone call, text, or email they’re responding to is actually official.

BBB Senior Director of Communications Jason Meza, speaking to KSAT news, explained that the IRS “has seen a record number of calls come in.”

“We’re seeing a lot more calls, emerging calls of AI, impersonating the IRS, impersonating collectors, tax collectors, debt collectors trying to get you to pay up or face penalties or jail time.” -Meza

IRS Typically Uses Snail Mail

How can you know what’s a scam and what’s not?

For starters, you can reject a phone call, text or email that claims to be from the IRS. According to Meza, the IRS “traditionally does not use these services. In fact, they will use snail mail.”

If anything arrives unsolicited and you haven’t first reached out to the IRS for information, be very cautious.

Beware Unexpected Tax Refund Offers

Another tip? Common tax scams in 2026 include messages claiming that the target qualifies for new tax credits or larger refunds — relying on the target’s hope for more money or a better deal to trick them into sharing personal information.

Clicking on a link from these scam messages will direct victims to offer up Social Security numbers and birth dates, paving the way for full-blown identity theft.

If you think you’ve been a victim of a scam like any of the ones mentioned in this article, you should report it to the BBB, the IRS, and/or the Federal Trade Commission.

Microsoft has released new updates for its Copilot AI offering, including Copilot Cowork, which can plan, coordinate, and access various tools to complete tasks with human supervision.

Likewise, Microsoft announced upgrades to Copilot’s Researcher agent. With “Critique,” users can use and compare outputs from different AI models, including OpenAI’s GPT and Anthropic’s Claude.

These upgrades signal a shift within enterprise AI offerings that focus more on solving multi-step, complex workflows and granting more autonomy to AI systems.

Microsoft Releases Copilot Cowork for Early Access

On Monday, Microsoft unveiled new upgrades to Copilot, including a new agentic AI tool known as Copilot Cowork.

Cowork behaves autonomously in order to complete tasks set out by the user. Once it has its goal, with human oversight, it generates a plan, coordinates tasks using the various tools and files it has access to, and presents its findings.

Currently, Cowork is only available via early access for participants in Microsoft’s Frontier program.

Copilot Researcher Agent Gets GPT and Claude Upgrades

As well as Copilot Cowork, Microsoft rolled out a series of upgrades to Copilot’s Researcher agent. One new feature, “Critique,” sets up a system where one AI model can evaluate the responses of another.

OpenAI’s GPT generates the response, while Claude reviews it for accuracy and quality, before it’s presented to the user. Microsoft hopes to invert this workflow in the future, where Claude generates and GPT reviews.

Likewise, the “Council” feature allows users to pull outputs from different AI models for the same request, allowing direct comparison between responses.

In a post on LinkedIn, Microsoft CEO Satya Nadella wrote: “You can run multiple models on the same prompt at the same time, so you can see where they align and diverge, and understand what each adds.”

Enterprise Shift to Multi-Step Workflows

These new developments from Microsoft represent a shift in enterprise-level AI offerings, which are more focused on multi-model AI systems that can be built into long-running and complex workflows. Businesses are now searching for ways AI can become embedded into their workflows, completing tasks autonomously.

In a blog post, Jared Spataro, chief marketing officer for AI at Work at Microsoft, described Copilot Cowork in particular as a shift toward AI that carries out connected sequences of actions, rather than simply acting as a response generator for loose, individual user requests.

Multi-model AI workflows are highly beneficial for businesses, especially as AI is increasingly used as a way to manage multiple systems, departments, and data sources. For example, they can be used to enhance consistency across projects that require input from several teams, such as resolving difficult customer queries.

Voice phishing is getting better and better: It’s now the second most popular initial entry point for cyber attacks, according to Google.

The Google Threat Intelligence Group’s latest report is out from incident response unit Mandiant, and it found that “highly interactive” voice fakery now makes up 11% of vectors observed across 2025.

That’s almost twice as frequent as email phishing attacks (6%), which were once a huge staple of hackers everywhere.

Voice Phishing Surged Across 2025, New Data Shows

“Highly interactive voice phishing saw a significant surge to 11%, becoming the second-most commonly observed vector” in 2025, the report said.

However, it notes that “exploits remained the most common initial infection vector for the sixth consecutive year, accounting for 32% of intrusions.”

What does voice phishing look like in practice? Some groups target IT help desks with the aim of fully bypassing typical security measures like multifactor authentication (MFA) in order to get the initial access to software-as-a-service (SaaS) environments that they need.

Email Phishing Attempts Were Less Frequent

In a sharp contrast, email phishing fell to just 6% of intrusions last year. Social engineers appear to be moving to audio and away from text.

Other takeaways from the new report:

• Cybercriminals a lot faster at handing off initial access in targeted environments: Their window was just 22 seconds in 2025 – huge drop from eight hours in 2022.
• Global median dwell time is 14 days, up from 11 days. This “likely reflects growing sophistication, particularly in evading defenses.”
• Organizations are getting better at spotting threats internally: “52% of the time organizations first detected evidence of malicious activity internally, an increase from 43% in 2024.”

Cyber Criminals Love AI Tools

AI-powered voice phishing is just the start: Once they’ve gained access, hackers can harvest OAuth tokens and session cookies or steal hard-coded keys and personal access tokens. They can then use all that data to infiltrate downstream environments and steal mass amounts of customer data.

The new report joins similarly troubling analyses from recent months.

Most recently, we covered a study by Armis that claimed cyber crime had hit a global “tipping point.” We’ve also heard from yet another study which determined that a full 68% of professionals say existing security tools can’t mitigate new threats.

50% of all working hours are set to be undertaken by AI agents in the near-future, says a new report from Accenture and The Wharton School. The findings appear in The Age of Co-Intelligence, which examines how human and AI collaboration is evolving.

Elsewhere, the study concludes that so-called “minimum viable” AI-led organizations will require at least 60 enterprise agents to support key functions such as R&D, customer service, and finance. 35 of these agents set to be digital, and the remaining 25 will be physical.

The research is part of a growing body of evidence that, increasingly, AI is no longer just used as an enabler, but as a tangible replacement for human labor. While businesses are pushing back on this narrative, it is becoming harder to deny.

Half of All Working Hours to be Conducted by AI Agents

50% of all working hours will be handled by AI agents in the near-future, according to new research from Accenture and The Wharton School. The consultancy giant itself laid off thousands of employees last year in what many saw as a pivot towards AI.

The study looks at how the relationship between humans and AI has evolved, contending that usage has shifted from “simple augmentation” to “co-intelligence.”

To gather findings, the researchers analyzed task-level data across 18 industries using data from O*NET (Occupational Information Network) and the Bureau of Labor Statistics, corresponding to more than 120 million workers across 18 industries.

Findings Illuminate Changing Human-AI Relationship

The report argues that organizations will soon require at least 60 agents to support important business functions, comprising 35 digital agents and 25 physical ones.

Consumer goods and services businesses will see the highest percentage of work undertaken by these agents, with 58% of working hours to be automated (29% digital agents and 29% physical ones).

At the other end of the spectrum, healthcare businesses will remain largely untouched by AI automation, with 28% of hours impacted by digital agents and just 13% impacted by physical ones.

While it’s thought that the emerging dynamic will have a positive impact on business output, the researchers issued a note of caution: “Humans must stay in the lead by setting direction, defining guardrails, challenging analysis, making trade-offs, and owning outcomes.”

Growing Disconnect Between Employee Fears and Business Assurances

The Accenture-Wharton report is the latest in a long line of similar findings that posits AI is no longer just a tool for augmentation, but poses a genuine threat to jobs everywhere. Recently, it was revealed that a staggering 54% of companies plan compensation cuts due to AI, with a further 26% set to lay off staff in favor of automation entirely.

Increasingly, workers are fearful that their job security is at risk due to increasing investment in the nascent technology. Several high-profile companies have already laid off staff while increasing their AI budgets, including Accenture, Salesforce, and Microsoft.

A number of companies have also walked back their AI pivots, with a rift emerging between employee concerns and the language that their bosses are deploying with regards to AI. It’s expected that this division will only continue to deepen as AI uptake increases.

37% of consumers trust brands more when their email communications feel human, according to a recent study.

While consumers aren’t set against businesses using AI in brand emails, above all, the study shows that personalization and relevance are the key makers of trust.

Businesses can utilize AI tools for monitoring email marketing campaign success, however, the human touch remains a vital component to ensure communications feel personal.

Consumers Trust Brand Emails That Feel Human, Despite Being Less Polished

In a recent Adobe Express study, 37% of consumers revealed that they trust brands more when marketing emails feel human, even if this makes them appear less polished. This suggests that personal and authentic communications matter more to consumers today than perfection.

Similarly, 18% of consumers said they have unsubscribed from marketing emails in the past because they suspected it was written by AI.

Adobe Express surveyed 1,007 consumers in the US as part of its study, with data being collected in December 2025.

Consumers are Generally Wary of AI in Email Marketing

On average, consumers estimated that they receive around six AI-generated emails per week, highlighting an increased use of AI within email communications. However, this isn’t always welcomed, as almost half (46%) of consumers surveyed said they would be more likely to unsubscribe from emails if they knew an email was written by AI.

Despite not all marketers embracing AI in this space, these findings suggest that consumers are increasingly paying attention to the way emails are written, and that AI use can be detrimental to how much trust is fostered between brands and their audiences.

Additionally, 78% of consumers highlighted that messages that felt too salesy or pushy made emails less appealing, as well as wordy emails (46%) and generic messaging (37%).

The Way Forward Blends AI with Human Creativity

Overall, 60% of consumers said an email that sounds like a real person wrote it is what makes it feel truly personalized.

That being said, 37% of consumers said they don’t care whether a marketing email sounds human or AI-generated, so long as the content within is useful and relevant to them. This positions personalization and relevance as the gold standard of marketing emails.

These findings state that consumers are not entirely averse to the use of AI in email marketing, however, businesses should focus on creating a balance that includes both human creativity and AI, in order to achieve the personalization and relevance consumers feel from a human-written email.

AI can be highly beneficial to campaigns, by monitoring key metrics and analyzing trends for example, but businesses seem apt to lose their audience when they attempt to replace the trust consumers feel from human creativity.