---

The Trump Administration’s handling of communication on military strikes in Yemen . Monday’s bombshell report that top government officials—including Defense Secretary Pete Hegseth, National Security Adviser Michael Waltz, Vice President JD Vance, Secretary of State Marco Rubio and Director of National Intelligence Tulsi Gabbard—used commercially available messaging app Signal to discuss airstrikes against the Houthis earlier this month shocked the cybersecurity community, as well as the general public. But what may be more shocking is the way the story got out: Waltz inadvertently added The Atlantic’s editor-in-chief and longtime national security journalist Jeffrey Goldberg to the group chat.

An anonymous official at the Cybersecurity and Infrastructure Security Agency, which is the Department of Homeland Security’s digital safety arm, told Forbes’ Thomas Brewster that the situation is “absolutely ridiculous on so many levels.” Discussing classified information, like war plans, on a platform that is not secure The official described it this way: “Individuals in this administration think they are playing James Bond, but they have the skills of a Dr. Evil.”

As far as messaging apps go, Signal is one of the most secure. Messages are encrypted, but given that it runs on fundamentally insecure devices—personal phones and computers, both of which are vulnerable to hackers—. In a Washington Post article, current and former national security officials outlined the way that the federal government typically requires classified information, like war plans, to be discussed: In person, with personal devices left outside of the room, or through pre-approved, secure communications devices that key officials can access 24/7. Another problem with Signal: Anybody can download and use it, adding to the possibility of inadvertently adding someone—like Goldberg—to a conversation in which they do not belong.

So far, . He’s defended Waltz, who admits to adding Goldberg to the chat. The president also said the information discussed was not really classified, though many national security experts, as well as ranking Republicans, have disagreed with him. A YouGov poll this week found 53% of people think this represents a “very serious” problem, and nearly half think officials broke the law. The top Republican on the Senate Armed Services Committee called on the Defense Department’s inspector general for an expedited investigation into the texts. A government watchdog group has also filed a lawsuit over the way the messaging app would have deleted all messages.

While the vast majority of businesses don’t handle classified information, : Employees use devices, apps and programs that are not designed to protect proprietary data for work duties. Forbes contributor Zak Doffman writes about why this is a temptation for employees—and why it is a problem. A CyberArk study released late last year found that 80% of employees use personal devices to access workplace applications, and 52% share confidential workplace information with external parties. It seems everyone can use a refresher on the best practices for workplace data.

In recent months, this decade. I talked to Liz Durst, who built IBM’s Qiskit quantum developer tool and is now VP of Riverlane’s QEC Community, about the development of this area and what CIOs should know. An excerpt from our conversation is later in this newsletter.

Anthropic is teaming up with Databricks to give companies the . The two companies announced the five-year partnership, worth $100 million, this week, writes Forbes’ Richard Nieva. Since many companies already use Databricks’ storage and analytics program to store their enterprise data, the partnership will allow for use of Anthropic’s powerful Claude 3.7 Sonnet model—which can understand conventional language and also “reason” through multi-step problems—on top of that. “This is the future of enterprise AI,” Databricks cofounder and CEO Ali Ghodsi said in a press release. Anthropic Chief Product Officer Mike Krieger told Nieva that while this partnership will help Databricks customers, its feedback and performance channels will also inform Anthropic on how to make and deploy better AI agents.

When Chinese company DeepSeek initially launched its R1 model in January, the AI world took notice, and U.S. tech stocks dropped as investor confidence in more expensive platforms was rattled. This week, but there’s been much less chatter and residual effects. According to the company, the upgrade to its V3 LLM, called DeepSeek-V3-0324, includes “significant improvements” across several benchmark tests for language, as well as upgrades to its front end, as well as Chinese writing proficiency and search.

, with a few share prices slipping after it was released Tuesday morning. (Nvidia’s stock has seen a much steeper drop this week, but it likely has more to do with anticipated fallout from tariffs from the Trump Administration taking effect next week.) Apple CEO Tim Cook, visiting China this week for a company developer conference in Shanghai, praised DeepSeek, telling the China News Service that the company’s models are “excellent,” according to the South China Morning Post.

, but payments to stop the attacks are down. Forbes senior contributor Davey Winder writes that AI-powered cybersecurity protection firm Ontinue’s most recent threat intelligence report showed ransomware attacks up 132%, but payments were 35% lower—meaning both companies were less likely to pay, and cybercriminals could be looking for data and other information when they attack a company. Ransomware isn’t the only cybercrime that’s seeing attacks surge. AI technology is powering an attack known as vishing—voice phishing, in which a deepfake of a trusted person’s voice is used to trick people into handing over sensitive information or authorize fraudulent transactions. So far in the first quarter of 2025, vishing-related incidents are up 1,633%, Ontinue found.

The last several months have brought huge developments to quantum computing. Last month, Microsoft unveiled a new matter-shifting quantum chip that it says opens the path to developing a fault-tolerant quantum computer within years, while Google unveiled its Willow quantum chip in December, which could solve a problem in five minutes that would theoretically take a supercomputer longer to solve than the age of the entire universe. Riverlane, a company committed to making quantum useful through quantum error correction, recently hired Liz Durst—who built IBM’s Qiskit developer tool—as its VP of QEC Community. I talked to Durst about the state of quantum computing and what CIOs should know about it. This conversation has been edited for length, clarity and continuity.

It’s just getting started. We talk about the QEC [quantum error correction] community. We’re talking about 300 to 500 people in the world. This summer, there’s a QEC conference that brings together all of the leading experts in this field, and they expect to get 400 people attending this. That’s the number of people that are deeply capable of understanding the theory behind these systems, let alone how to go and engineer that into this future view of a fault-tolerant quantum computer. Right now, there exists very few software tools, mostly open source by the Google team, providing early software tooling that technology teams need to start modeling.

We need to go from there being 300 to 500 of these people to every single quantum computing provider and technologist working with quantum computing to skill up and understand how to shift from all the things we did to work with the near term quantum computers of the [noisy intermediate scale quantum] era. We’re now suddenly entering the QEC era. It’s this huge problem and why Riverlane is investing in putting all of this open tooling out into the ecosystem so that as we start to enable it, we can actually grow it. If there’s not enough people to solve the problem, we’re going to be stuck. Now, it’s such a pivotal time to arm the full community with these useful tools to move in that direction.

With quantum, it is such a niche field, but when combining usefulness with amazing education and learning resources, suddenly this community is doing this iterative thing of learning and building and learning and building, and it all feeds itself. That’s for sure what we saw with the Qiskit community—this giant open source community where we have people going and building and starting to develop algorithms to use on quantum computers. I think the same thing is going to happen for QEC.

If you look across the entire industry, between 2028 and 2029 is when you’re going to start seeing a MegaQuOp, which sounds like a lot of letters and acronyms, but it’s 1 million error free operations—where we think you’re going to hit the threshold of there being a useful quantum computer.

But the more controversial part is that [people will need to] think of those early demonstrations as scientific computing. This isn’t like suddenly it’s ready to get embedded into your laptop and it’s ready to go. It’s going to be a long haul. And you’re going to see this rush: Now we’ve proved quantum computing is the real deal. We now need to start a much more aggressive strategy to get these systems ready for production and embedded into all of these different applications.

There’s quantum computing itself, and then there’s quantum-safe or quantum-inspired computing. Right now, people are looking at quantum-safe cryptography. NIST is publishing standards [so companies can] think about writing encryption in this way, so that in the next five-plus years when quantum computers are around, you have quantum-proofed your systems to be able to operate that information architecture.

I think the much longer term challenge is as soon as [a useful quantum system] happens, those who have institutional knowledge [will be in demand]. It takes so much operationally, standing up small innovation teams starting to explore near-term innovation, and [figuring out] what are the useful applications for quantum computing. You see this happening in finance, chemistry, high energy physics, pharmaceuticals. I think it’s a worthwhile investment to start now so that you are a part of it.

This is a very new type of computing that may not necessarily move your bottom line immediately, but if you’re not prepared for it, it could completely take you by surprise. I think it sort of reflects the shift you see now with AI: What do we do now? We maybe need to shift something substantially, but we hadn’t been exploring it as seriously, until suddenly the big shift is happening.

It’s not going to be true for everyone, but I think it’s worth it to do an assessment: Where is your business positioned in terms of quantum readiness? There are a whole number of consultancies that are willing to help you figure that out. But in my opinion, it’s absolutely worth at least trying to skill up some of your workforce on what is quantum computing and where might this be an impact.

The United Arab Emirates pledged an —following a meeting between Trump and UAE National Security Adviser Sheikh Tanoon bin Zayed Al Nahyan last week.

Amount of the investment over the next decade

Amount of a separate “Stargate” AI infrastructure investment announced in January that MGX, an investment firm that Sheikh Tanoon chairs, is part of

One of the opportunities that Sheikh Tanoon highlighted as a way for the U.S. and UAE to strengthen their partnership and unlock new avenues, in a meeting with Treasury Secretary Scott Bessent, according to a press release from the UAE Embassy

. The future of data engineering requires humans and machines to work together.

Leaders don’t have to be loud. Here’s how to be powerful, command attention and make a difference as a soft-spoken leader.

QUIZ

Austrian nonprofit privacy group None Of Your Business is suing OpenAI after a man asked ChatGPT for information about himself and received an inaccurate response. What did ChatGPT say the man had done?

A. Taught the 9/11 hijackers at flight school

B. Murdered his children

C. Led the LockBit ransomware gang

D. Plotted to kill Pope Francis

See if you got the answer right here.