The most recent information for chief technology officers and other technology-focused leaders is available in Forbes ‘ CIO newsletter, which has been published. to get it delivered to your inbox every Thursday.
President Donald Trump has spent his first weeks in office making sweeping adjustments to the federal government, only as he promised on the campaign trail. In some areas, there is currently a heated debate over whether those changes are fair, legal, or in the world’s best interest. Discussion over the security risks that come with the Trump Administration’s techniques, however, has been much more subdued.
The Elon Musk-led Department of Government Efficiency’s method of accessing federal data has raised alarms among analysts and media reports everywhere. Musk and his associates allegedly forced their way into accessing sensitive data centres, including the national payment system used by the Treasury Department, by reportedly being fresh programmers with connections to networks and comments on social media that make them unlikely to pass background checks.
For accessibility, which has been protested by profession government employees and Democratic leaders, has been the subject of intense litigation, with courts now ruling the group cannot access the payments system and limiting Depends to just read-only access ( though at least one DOGE worker falsely had the ability to make changes to the payment database for a short time ).
A prosecutor put the release of Musk’s team’s request for access to at the Department of Education, which includes sensitive information about the identities and incomes of millions of people who have borrowed money to pay for higher education earlier this year. Apparently, Expand team members are using AI software to analyze inconsistencies in the Azure platform and run education grant data. And the original Tesla expert now leading the government’s Technology Transformation Services has asked for privileged access to at least 19 internal government systems, including data collection Data. Federal employment table USA, www.gov.gov. authorities text alert system Contact and the Go.gov. state, Forbes ‘ Cyrus Farivar wrote.
According to Tony Bradley, a top contribution to Forbes, DOGE is a security crisis that is permeating the country right now, and the level of entry sought reflects the kinds of attacks carried out by foreign nations against the United States. It’s unclear where the data being accessed is going and how it’s being used, in addition to possibly breaking a number of client data protection laws. Even if data is only accessed by skilled workers with appropriate certifications, more ways to access information equals more ways security can be breached.
There are no details about DOGE’s data entry or the background of those responsible for the work. If the stories about AI are accurate, sensitive government information could also be made public, incorporated into” training information” for a larger AI software, and potentially shared with people outside the state. But even if no larger security incidents come out of this entry, it’s a breaking of government confidence. People inherently have faith in the state to protect their personal information, which does not seem to be the case at this time.
The R1 type from Chinese AI business DeepSeek appeared to have sparked an explosion in the AI market when it first appeared. But in reality, people in technology explored it, learned its teachings and continued to work toward improving their own models. What does DeepSeek mean for CIOs, in the perspective of Adnan Masood, the CEO of the technical change company UST. Later in this publication, an excerpt from our conversation is included.
ARTIFICIAL INTELLIGENCE
Elon Musk is currently doing more than just looking through sizable federal records. He also submitted an uninitiated bid this week to acquire OpenAI’s volunteer shoulder for$ 97.4 billion. OpenAI CEO Sam with a mocking post on X, the Musk-owned social media network formerly known as Twitter:” no thank you but we will get tweets for$ 9.74 billion if you want”. The price is related to Musk’s$ 44 billion purchase of the social network in 2022, which has declined.
Although Musk and Altman have been at odds with one another for years ( Musk was one of OpenAI’s cofounders and has sued the business for their plans to turn into a for-profit organization ), this action could be much more devastating. Forbes ‘ Richard Nieva writes that Musk’s bid good complicates things for Altman. Musk’s charge gives the volunteer controlling OpenAI a floor value because Altman is attempting to convert the business to a for-profit organization. And as of right now, anything less would be challenging for board people to recognize. Musk tacitly admitted his game plan, saying in a court filing he would withdraw his$ 97.4 billion bid if OpenAI stays a nonprofit.
No other reliable figures have been released for the nonprofit division of the business, which is likely much less than Musk’s proposal, which was previously valued at$ 40 billion by The Information. According to Nieva, the entire business will be fair$ 260 billion following a$ 40 billion funding from Softbank.
POLICY + Laws
Vice President JD Vance addresses the Paris-based Artificial Intelligence Action Summit.
LUDOVIC MARIN/AFP via Getty Images
One of the accepted beliefs in technology is that in today’s global economy, individuals from all over the globe use the same systems and online. Any company or program that wants to conduct business in that area must agree with new rules if one government entity passes them. The EU is currently home to the most stringent regulations, and many American-based businesses actively adhere to the coalition’s guidelines for data protection, privacy, and content moderation. But at this week’s Artificial Intelligence Action Summit in Paris, Vice President JD Vance against the bloc’s digital regulations, urging them not to overregulate AI because the U. S. intends to be the leader and “gold standard” on the technology.
According to Vance,” the Trump Administration is alarmed by reports that some foreign governments are considering tightening the screws on U.S. tech companies with international footprints.” America” cannot and will not accept that,” and we believe it to be a terrible error. At this moment, we face the extraordinary prospect of a new industrial revolution… However, it won’t ever happen if excessive regulation prevents innovators from taking the risks that are necessary to advance the ball.
Trump has yet to develop comprehensive AI laws, but one of his first priorities following taking the oath was rescinding former president Joe Biden’s 2023 executive order on AI development, which aimed to regulate AI development. Biden’s order focused on the government protecting users ‘ privacy and security, keeping bias out of AI systems, and controlling systems that could potentially bring risks to humanity. Trump’s actions will “unshackle America’s innovation economy,” according to the lobbying group NetChoice.
However, the EU does have laws in place that regulate AI. The EU AI Act is new, and mostly around where AI cannot be used, prohibits exploitation or abuse of biometric data, and institutes some safety-related regulations around high-risk systems. Where Trump’s policies or any laws passed by the Republican-controlled Congress will fit into that category is still to be seen.
CYBERSECURITY
Ming Yeung/Getty Images
Last week, a Washington Post report said U. K. security forces secretly demanded Apple create a way for them to retrieve all of the encrypted content any user worldwide has uploaded to the cloud. According to Emma Woollacott, a senior writer for Forbes, Apple added optional end-to-end encryption to this data in 2022, making it unattainable for Apple to access. According to Woollacott, a contentious British law passed in 2016 makes it possible for the security forces to demand it. Referred to by critics as the” Snooper’s Charter”, Information Technology and Innovation Foundation Vice President Daniel Castro told Woollacott this law weakens consumer data protections—and can also justify authoritarian regimes that want to bypass encryption on personal data.
Apple and the British government have not made any comments on the matter, but that considering Apple’s previous positions on data privacy and protection, the company may stop providing encrypted storage in the U.K., which could make it in base compliance with the order.
BITS + BYTES
What AI Professionals Want You To Think About DeepSeek
UST Chief AI Officer Adnan Masood.
Carver Mostardi
While investors, analysts and policymakers reacted to last month’s introduction of Chinese AI chatbot DeepSeek with panic, tech transformation company UST’s Chief AI Officer Adnan Masood said DeepSeek shows new applications and approaches to old tech, but it doesn’t necessarily bring a dramatic change to the landscape and is a victory for open source development. He explained what DeepSeek actually means for the tech sector, and what CIOs ought to be aware of. For clarity, consistency, and length, this conversation has been edited.
Was the large upheaval caused by DeepSeek’s introduction merited?
Masood: The way they’ve put together a pipeline has done some amazing work. People have seen how we used to have this$ 1 billion price tag, and now that you have$ 6 million, you can get the training paradigm with almost the same level of accuracy, if not greater. So that’s a big shock, and that automatically in peoples ‘ minds translated to: Why are we spending so much money into buying GPUs? There are suggestions that we’ll be able to run these things on local GPUs because of the reduced scale.
However, it’s a very simplistic story, and there’s nuance to it, like everything else.
If you want to right now run a model like DeepSeek R1, it requires about 400 gig of video RAM. You don’t have 400 gig of RAM for your computer. You have 12 gig of RAM for video. You need to get those GPUs in order to be able to inference and run it, so it’s not the end of the GPU as we think. Additionally, you must be able to introduce it to the general public. Additionally, that calls for GPUs. It’s not that the GPU market has gone completely down. Simply put, it has excellent optimization. It represents a victory for open source, in my opinion. A victory of this ability to bootstrap and build things when they didn’t have all the resources to make it really well.
Overall, it actually gives more incentive for labs with bases in the United States to develop more efficient, less expensive, faster, and better models.
Is DeepSeek something the tech community can emulate and learn from?
That’s exactly how researchers, scientists and AI practitioners look at it. Yann LeCun, the chief AI scientist at Meta, has made it clear and consistent that open source and open science are important. Lama was made open source, which made it possible for many businesses. I think 2024 was really the era of democratization of AI: When AI became mainstream, and people knew that they had access to these models. They are able to run on an enterprise level and have access to both self-hosted and hosted models.
I believe that the era of agentic AI will come in 2025, powered by both small and large language models. The sentiment is shared between researchers. No one wants closed-offering science. It’s beneficial to have open and available science and research, as long as value propositions, security, compliance, and laws and regulations are followed within that dynamic.
We are seeing now the whole idea of” the Wall”, where models are not improving the way they were supposed to improve. That notion of having that Wall is completely destroyed by this. No wall exists. This shows you how, with low cost and impressive performance, you can get]progress]. Additionally, you can improve and do distillation. You must address issues involving data security, regulatory complexity, geopolitical risk, and many others.
What lessons should CIOs take from the emergence of DeepSeek and how can they apply it to their businesses?
There is always a good strategy for doing things. One of the most important lessons they can take away is the importance of distillation, cost savings, and the use of expert models. We always thought]primarily about ] the GPU, the way we train transformers, and this kind of change shifts that.
AI will have these directional shifts if it is combined with expert models. It is extremely important to be able to pivot and rely on the enterprise ecosystem and the scale economies. There’s a status quo and there’ll be disruption, and I think DeepSeek really poses for CIOs a genuine risk of disruption to large closed-source AI players. You probably want to reevaluate your AI strategy to be able to pivot if you are getting on board with the closed source adoption. It demonstrates how dynamic the market is right now and how it is evolving.
Also, it makes people think more about AI ethics: ethical AI, responsible AI, accountability. What kinds of rules will be in place? How do you interact with potential adversaries? What are you doing around sovereignty, Trojan horses or reliance on a foreign AI technology? It raises a lot of strategic concerns for CIOs.
FACTS + SUGGESTIONS
As AI use grows, hundreds of billions of dollars are on building and outfitting AI data centers this year.
290: The number of data centers that East Daley Analytics energy consultant Zach Krause said are likely to be constructed in the near future
By 2030, a total of 81 gigawatts would be required to run all of the new data centers, roughly equivalent to the energy used by Texas.
‘ We will see its use skyrocket, turning it into a commodity we just can’t get enough of’: Microsoft CEO Satya Nadella posted on X about the rise in use of AI as it gets more efficient and accessible
STRATEGIES + ADVICE
Continuous threat exposure management is a new approach to better preparing you for cyberattacks. Here’s on how to use it.
Storytelling can improve your communication skills and increase your speaking power. Here are some methods to make it work for you.
VIDEO
QUIZ
The House Ethics Committee did something unconventional to its website in December. What was it?
A. Briefly added a countdown clock to President Trump’s inauguration at the bottom of the page
B. Give more information about the committee’s activities a large picture of a Christmas tree.
C. Added code so it wouldn’t be indexed by search engines
D. Removed the committee’s and staff’s general contact information.
See if this is the solution you need.