AI and quantum technologies present both opportunity and risk: they can dramatically improve companies' capabilities and accuracy, but they also create an entirely new, expansive threat landscape. Even if your company is in a less technical industry, these threats are still present—and have the potential for massive harm to your business and customers.I spoke with Andrew McLaughlin, chief operating officer of SandboxAQ, about how CIOs should face and address these threats. And while companies are very different, the answer to dealing with the threat is the same: know what’s happening on your network. An excerpt from our conversation is later in this newsletter. Until next time.This is the published version of Forbes’ CIO newsletter, which offers the latest news for chief innovation officers and other technology-focused leaders. Click here to get it delivered to your inbox every Thursday.FROM THE HEADLINESPhoto Illustration by Sheldon Cooper/SOPA Images/LightRocket via Getty ImagesOpenAI has been practically synonymous with the generative AI revolution in the U.S., but it could be in trouble. The Wall Street Journal reported that the company missed its own targets for new users and revenue, raising concerns from its CFO that it may not be able to pay for future computing contracts without faster revenue growth. OpenAI has racked up several infrastructure deals in the last year, reportedly planning to spend $600 billion on compute by 2030.The report hit the tech sector like a brick, and calls into question the company’s long-talked-about plans to IPO later this year. OpenAI disputes the report, with spokesperson Steve Sharpe telling Forbes that the Journal’s story was just “clickbait.” However, when the story broke earlier this week, it brought stock down for some of OpenAI’s largest investors and partners, including Broadcom, Oracle and SoftBank. The company that would be hit hardest, though, is CoreWeave. Forbes’ Phoebe Liu and Richard Nieva write that OpenAI makes up about a third of CoreWeave’s planned revenue over the next seven years. And while OpenAI isn’t the AI cloud company’s only big customer, a failure to make payments will set CoreWeave’s business off balance. Infrastructure costs have been the biggest question mark for the AI space—projections of more spending in this area than anticipated have been enough to drop Big Tech stock prices. But right now, analysts told Forbes, the whole financial equation of AI is also a pretty big question.ARTIFICIAL INTELLIGENCEThe blockbuster trial between Elon Musk and OpenAI started this week. The world’s richest man took the stand to accuse Sam Altman—who cofounded OpenAI with Musk and other researchers in 2015—of being greedy and changing the company’s mission. OpenAI was initially founded as a nonprofit, but restructured as a for-profit entity last year, long after Musk left the company in 2018. The trial, expected to take place over the next month, could be seen as a case that might shape the destiny of the AI landscape. But it can also rightly be seen as a grudge match between two rival billionaires.The trial opened with two days’ worth of testimony from Musk, who said he “was a fool” to fund OpenAI to the tune of $38 million—free funding that helped launch a company now worth about $800 billion. Musk accused OpenAI of trying to “have its cake and eat it too,” by switching from an enterprise with “some moral high ground” to one based on profits.OpenAI has not yet had the chance to present its defense, but the New York Times reported the company’s lead counsel, William Savitt, said in his opening statement that the case was all “sour grapes.” “My clients had the nerve to go on and succeed without him,” Savitt said. “Mr. Musk did not like that.”OpenAI’s nonprofit foundation still has a 26% stake in the company, which Forbes contributor Paulo Carvão writes is worth about $130 billion. It has already committed funding to health breakthroughs through open sourced datasets and practical technical solutions for AI resilience, and has the potential to rewrite the corporate giving playbook—one of the things Carvão writes also seems to be on trial.BIG DEALSThe exclusive arrangement between Microsoft and OpenAI has been renegotiated, and OpenAI can now license its models and products for any third party and any cloud—including Microsoft competitors Amazon or Google. Microsoft will no longer pay OpenAI a revenue share, and OpenAI’s revenue share payments to Microsoft now have a cap.The deal between the two tech powerhouses, initially struck in 2019, was significantly renegotiated last year as OpenAI sought other providers to meet its infrastructure needs. This further streamlines the partnership, and the companies said in a statement that it “strengthens our joint ability to build and operate AI platforms at scale while providing both companies the flexibility to pursue new opportunities.” In practice, it could bring OpenAI’s products to different enterprise providers and give Microsoft customers more choices.BITS AND BYTESThe New CIO Security Playbook: Why Next-Level Visibility is the Only Defense Against Shadow AISandboxAQ Chief Operating Officer Andrew McLaughlin.SandboxAQ, GettyThere are many possibilities—both good and bad—at the point in computing where AI and quantum meet. And while there are many companies dedicated to providing the next generation of solutions—including SandboxAQ, which began as a spinoff from Google and is currently developing cutting-edge B2B platforms for a variety of industries and uses—there are also CIOs of non-tech companies just trying to keep up with cybersecurity threats. I spoke with SandboxAQ Chief Operating Officer Andrew McLaughlin—who previously served in a variety of tech positions including deputy U.S. CTO under President Barack Obama, head of global public policy at Google and president and CEO at Assembly OSM—about what today’s CIO security playbook should be. This conversation has been edited for length, clarity and continuity.There are so many threats out there today, and CIOs are aware of the dangers of all of them. They also most likely know where the security gaps are in their enterprises. What should they be doing?McLaughlin: The playbook is not all that mysterious or hard. The first thing you do is get visibility. Once you have the ability to detect where everybody’s using AI, then you apply controls, limits, set policies, that sort of thing. But we’re at the stage right now where most CIOs are like, ‘How do I get the visibility? How do I even know?’ Sadly, that’s a whole new tool suite. You’re trying to stop spending money, stop adding different tools. But the reality is you need some kind of radar for what’s happening on your network: what’s okay and what’s not okay.The CIO playbook is like, ‘Well, suck it up.’ You need to start instrumenting yourself to know what’s going on, see what the problems are, start fixing [them] and have some kind of control layer. The scary hard stuff is the external AIs that are coming at you. You can’t depend on perimeter defensing or you can’t erect walls. You have to be able to do the thing that they’ve been doing the last 10 years: every device has to have an ability to signal what’s happening and stop what you don’t know. That’s the playbook. You just have to take it seriously and invest money. Tell the CEO, ‘You’ve got to start budgeting for this.’There’s definitely been a rush to spend money on AI technology the last few years. Have most companies invested in this?It’s not that they’re dum-dums about it, it’s just that it’s moving very fast. You use Gemini and Copilot, you could run an experiment and see what you can get away with, right? The big hyperscalers have tools that the CIO or network administrator can use. That’s what makes shadow AI more interesting: That’s somebody using Gemini that’s not allowed to use Gemini, outside the orbit of the network administrator’s control. If it’s a Copilot shop and somebody’s using Gemini, tools don’t matter. We need to have something which is model agnostic. We need to have something only focused on defending you.In some ways, if you break this down, it’s not that different from problems that we’ve had with botnets. It’s just that they’re much faster, much more sophisticated. They’re tireless. The attacks coming from the outside just keep trying things, using their ability to overwhelm your defenses. The really scary demos I’ve seen are bits of code that get into your network because they’re perfectly benign, and then they have an ability to adapt and generate new capabilities after they’ve gotten inside.What’s the best way to go about an informational training session for employees that talks about the risks that unauthorized AI use can cause, but not talking up the benefits and tempting them?You go positive and negative. So deliver some capabilities that satisfy the need. I remember in driver’s ed, they would show these extremely gory movies of highway road accidents and stuff like that. And even though we made fun of it at the time, that actually really did stick with me. I was a much better driver having sat through those absurd movies about Sheriff Joe going down the highway and seeing the twisted bodies of dumb teenagers speeding into the pylon. I think this is a moment to scare people straight.COMINGS + GOINGSTelecommunications provider Telephone and Data Systems named Bill Case as its new senior vice president and chief information officer, effective April 27. Case joins the company from WOW! Internet, Cable & Phone where he worked as executive vice president and chief information officer.Tax compliance software firm Sovos appointed David McCann as its new chief technology officer. McCann most recently worked in the same role at Self Financial, and he has also worked in leadership for Abrigo and Accruent.Staffing and recruitment software company Bullhorn promoted Jason Heilman to chief innovation & labs officer. Heilman previously led the firm’s development and launch of its next-generation AI solution.STRATEGIES + ADVICEWhile enterprises are providing AI to employees to improve their work, studies show that many aren’t happy with the way the rollout is going. A survey from Roloff Consulting found more than seven in 10 said their companies’ AI strategies are either reactive or nonexistent. Here are the six biggest complaints about AI rollouts from the survey.Work seems to be getting increasingly stressful. To commemorate the last day of National Stress Awareness Month, here’s why interrupting your day with five-minute “micro-chillers” can be beneficial for your mental health, as well as those who work with you.QUIZA U.S. company’s bid to acquire Manus, a Chinese AI company, was blocked by China’s government this week. Which company is now trying to unwind its progress in the merger?A. AmazonB. MetaC. GoogleD. OracleSee if you got the answer right here.