Computerworld

In the battle over whether the US or China will lead the world in AI, the argument for the US side largely can be summed up in four words: USA good, China bad. In this worldview espoused by big US AI companies, the Trump administration, and many Republican members of Congress, AI born and bred in China is portrayed as a grave security risk. The Chinese government, they claim, will use Chinese AI models to steal information from businesses, individuals, and government agencies. They also claim that because Chinese AI complies with Chinese censorship rules, US companies that use these models will be forced to comply with the rules as well. Critics also say that if US AI is overtaken by Chinese AI, the US economy could be dealt a crushing blow, because the hundreds of billion dollars being invested by US AI companies will go down the tubes. Because of this, Big AI firms Anthropic and OpenAI are lobbying to ban Chinese open-source AI from the US. Not all AI companies and their customers agree with a ban. Microsoft, Meta, and AI chip maker Nvidia, among others, say there’s room for both US AI models and Chinese open-source models. What’s the fight really about? To understand, let’s first dig into the tech behind most Chinese AI. ## What’s the fuss about? The fight is primarily over what’s called lightweight open-source AI. “Lightweight” means the AI models don’t require multibillion-dollar data centers to train, build, and run them. They need comparatively modest processing power. That’s attractive for businesses, because lightweight models are less expensive to license and run than much American AI tech, notably OpenAI’s ChatGPT and Anthropic’s Claude. The other draw for businesses is that Chinese open-source AI tech can be used for free. And perhaps more important that is that the AI is typically what’s called “open weight,” which means businesses can tweak the technology for their own purposes. When a traditional American AI model is trained, people who train it must decide how much “weight” to give to every parameter in every element of training data. For example, if AI were being trained to tell the difference between an apple and an orange, a great deal of weight would be put on the color of the fruit and the appearance of its skin. Less weight would be put on the fruit’s shape or size. When a company buys American AI, they don’t know how much weight is given to each training parameter. They don’t even know what the parameters are. In open-weight AI, though, businesses have access to those parameters — and they can assign their own weights to them. This means they can customize the AI more easily for their particular needs in ways they can’t with traditional American AI models. (Note that the terms open weight and open source are often used interchangeably when talking about AI, although technically they differ. Some open-weight models withhold some information such as their source code and training data — information that should be made public for a model to be considered truly open source.) ## What Microsoft says about the Chinese models Microsoft, Nvidia, Google, Meta, and many smaller companies don’t agree that Chinese lightweight open-source AI should be banned. They claim that security and censorship issues can be easily solved. And they believe use of the models will help the United States economically, not hurt it. Microsoft CEO Satya Nadella posted on X: “Open-weight models are essential to a healthy AI ecosystem.” He added that the models will “strengthen American competitiveness and expand economic opportunity, while protecting national security.” Microsoft was also one of more than 200 corporate signers of a document outlining all the reasons they believe open-weight AI is important. Microsoft isn’t just talking about the importance of using Chinese lightweight open-source AI — it’s begun offering it to its customers and is looking to expand it even more. Right now, DeepSeek, the first Chinese lightweight open-source model to make its way to the US, is available via Azure. And Microsoft is testing offering DeepSeek to customers of its Copilot Cowork agentic AI product. Keep in mind that Microsoft’s embrace of Chinese lightweight open-source AI has less to do with virtuous thinking than with profit-making. Offering these products to its customers will please both old and new customers because of the tech’s benefits. The more Microsoft offers to its customers, the more money it will make. ## What happens next? It’s unclear what happens next with the Chinese AI products. OpenAI and Anthropic are allied with the Trump administration in wanting them blocked or limited in the US. Still, there’s an increasing groundswell backing back their use. Bill Gurley, a Silicon Valley venture capitalist, told the New York Times: “You have two factions fighting over this issue. There’s the people who want OpenAI and Anthropic to own everything, and then there’s everybody else, including customers.” My bet is that in short term, OpenAI and Anthropic will get what they want, because Trump has got their back. But after Trump leaves office, don’t be surprised if Chinese open-source AI has its day in the sun, backed by Microsoft.

Just because Apple now has AI, a new folding iPhone, and a newly minted CEO doesn’t mean the litigation between it and OpenAI has gone away. Apple now wants to force OpenAI to let it look at the hardware it has been building, according to a new report. ## A reasonable request? It seems a reasonable request, doesn’t it? After all, Apple’s argument is that OpenAI has been engaged in trade secret theft to help it design and develop its new hardware. OpenAI’s defense against these claims feel flimsy, at least to this reporter. They seem to coalesce around something like, “We don’t need your trade secrets because we’re making something brand new.” The problem with that defense is, contextually, that while on this stated mission to do something completely new, the company has hired around 400 former Apple staff so far, including its chief designers. And Apple thinks part of that process has been OpenAI, in whole or in part, working to exfiltrate its trade secrets. ## What Apple wants — and why With those facts as your guide, Apple’s request to Judge Edward J. Davila seems reasonable. It wants to take a look at what OpenAI is developing to ensure its trade secrets have not been abused in the process of designing that product. Apple argues that if it is forced to wait until the product is released, then it will be impossible to make its trade secrets confidential again, particularly as the defense seems to consist of that pinky promise that no Apple trade secrets have been harmed. Apple legal also argues that it cannot be fair to allow OpenAI to defend itself by alleging its unreleased and unseen product doesn’t contain any trade secrets without permitting Apple — and the court — to verify that. While Apple’s counsel doesn’t seem to have said it, you could paraphrase the request as Apple telling the genAI firm, “Let us see what you are building; if you have nothing to hide, you have nothing to fear.” Except, of course, that while building its defense, OpenAI is giving many of us the distinct impression that it may have something to fear. ## What may happen next Despite the merits of the argument, I think Apple’s request will not prevail, in part because the court may assess that if Apple takes a look at OpenAI’s homework it may compound the risk of IP theft. But that doesn’t mean Apple’s attempt will fail outright, as the compromise position is likely to be the appointment of a trusted, independent, third-party expert witness to take a look at what Apple’s competitor is making in Apple’s stead. There is precedent for this. That’s more or less what happened when Waymo pursued a similar case against Uber, or when AMCS litigated against Sinovel. There are nuances to all three cases that mean they aren’t perfectly aligned, but that does seem a logical next step to this layman. Of course, just because it’s logical doesn’t mean either party is going to like it, but OpenAI could conceivably even suggest such an approach as it seeks to buy itself time to build and release its still mythical hardware. That’s also why Apple wants a chance to look at documents pertaining to that hardware to make very certain it hasn’t infringed any of Apple’s own trade secrets. ## A side order of humble pie All the same, if this case does indeed turn out to be a scenario in which one company has been found with its hand in the cookie jar, then the best possible approach for the company with crumbs around its mouth is going to be damage control. That’s going to take a lot less war-war and a great deal more jaw-jaw. It’s also going to require the intake of a very, very large slice of humble pie. Right now, it seems to me that Apple isn’t talking, and OpenAI isn’t hungry enough to take that first bite. Not yet. The case, number 5:26-cv-07078, rolls on. Now please subscribe to my daily, human-curated Apple-related news headline feed at The Core, or follow me on BlueSky, LinkedIn, or Mastodon.

Microsoft on Monday added itself to a growing list of AI vendors pledging to try to control the behavior of its AI models. “AI should not exceed human control. Models should remain subordinate to humanity, subject to meaningful human oversight and control,” the company wrote in the draft version of its Humanist AI Code of Conduct, released Monday with an invitation to the public to provide feedback. “Humanist AI develops systems with clear purposes, evaluated against real-world impact, and rejects the race to produce an all-purpose superintelligence that could evade these safeguards,” Microsoft wrote. “We are building something fundamentally useful and safe even if that means compromising on ultimate generality, autonomy, or capability.” Microsoft’s comments are roughly in agreement with recent posts by various major AI vendors, including those from Anthropic and OpenAI, which were endorsed by Elon Musk, CEO of SpaceXAI and Tesla, but nowhere in its 37 page document is there any description of concrete action. However, to be fair, almost none of the other major AI players have been specific about how they would control future AI models either. ## Acknowledges issues A key problem with many vendor attempts to impose AI limits is that all of these companies have thus far been unable to stop AI agents from doing almost anything, given the agents’ ability and willingness to sidestep or ignore guardrails. Microsoft has described its worries about the technology in the past, both when it started to curtail AI efforts among its own employees and when it announced the formation of the team that created the Code of Conduct. Its current post acknowledged some of the difficulties involved in pushing AI development while limiting its abilities. “Both under- and over-caution represent failure modes with different types of consequences,” the company said. “Under-caution can clearly result in more direct harm, but over-caution may occur more often and therefore may need more frequent correction. This is where proportionality to the potential for harm and safety context are particularly important. Responses and actions should take into account the estimated severity and likelihood of potential harms and adjust responses and actions accordingly.” It noted, however, that the term “harm” is “broad and often context-dependent, and there are nuanced gradations in potential severity and likelihood. Microsoft AI (MAI) Model responses should be tailored to that context and to a wide range of harms.” ## Laudable goal, but lacks detail Analysts and consultants generally agreed that Microsoft’s stated goal is laudable, but the lack of specifics and verification mechanisms makes it difficult to take the post seriously. Thomas Randall, research director at Info-Tech Research Group, also said he spotted some apparent contradictions within the document. “[It] says MAI models will not assist in manufacturing or modifying weapons. Yet Microsoft offers OpenAI’s GPT-5.2 through Secret and Top Secret government clouds for defense and national security workloads,” Randall said, though he acknowledged that this is not technically a breach of the Code because GPT-5.2 is not an MAI model. “The most meaningful parts of the Code of Conduct may exclude other parts of Microsoft’s actual AI business operations. Tensions like these appear in other forms throughout the Code.” But he added that Microsoft’s position is bolstered by its earlier Frontier Governance Framework that “provides pre- and post-training evaluations, six-month reassessments, third-party testing, phased releases and a commitment to pause development or deployment where high risks cannot be mitigated.” However, he noted, “it is still Microsoft that defines the thresholds, selects the evaluators, determines whether residual risk is acceptable and gives its own executives the final deployment decision.” Randall said he would like to see Microsoft, as well as other major AI vendors, deliver more verifiable data points, such as those from independent evaluators given continuous access and freedom to publish findings about the vendor’s actions. He also would like to see independent board-level safety oversight with authority to block releases, protected whistleblowing, accessible monitoring, audit logs, kill switches, and mandatory reassessment after model changes. However, Justin Greis, CEO of consulting firm Acceligence, pointed out that Microsoft was candid about the many elements that are not yet in place. “The current models are not yet trained on the Code, the evaluation framework is still being developed, and Microsoft explicitly says written objectives alone cannot ensure alignment or guarantee present-day behavior,” Greis said. “That distinction matters. Publishing a constitution for AI is useful. Proving that the system actually follows the constitution, especially when models become increasingly agentic, is the hard part.” And consultant Brian Levine, executive director of FormerGov, said Microsoft deserved a little bit of credit for at least saying that model capabilities should be limited. “Microsoft explicitly says it will compromise on generality, autonomy, and capability to keep systems safe and under human control, and that it rejects the race to build an all-purpose superintelligence. Coming from a company of Microsoft’s size and ambition, that’s a notable thing to put in writing,” he said. “For years, the assumption was that the frontier labs would chase maximum capability and treat safety as a constraint to be managed. A document that says the opposite, that says usefulness and control come before ultimate capability, is worth paying attention to, regardless of what follows it.” He added: “The real test comes next and it’s verification: measurable standards, independent assurance and a way for outsiders to check the commitments against what’s actually shipping. That’s the natural progression and it’s the part the whole industry still has to build.” ## Not doing the hard part But others argued that Microsoft is merely doing the easy part, the marketing part and is deliberately not committing to doing the hard part. “Promising to give up capabilities is easy when those capabilities don’t yet exist,” said Noah Kenney, principal consultant at Digital 52. “The real test will come when Microsoft has a model ready to ship that would close a competitive gap and decides to hold it back.” Until then, he said, the promise of responsible AI costs Microsoft nothing. “Microsoft’s code of conduct reads like a marketing document written to reassure customers, regulators, and its own employees,” Kenney noted. “The problem is that no one knows how to make those promises specific, which leaves Microsoft asking for trust before they can explain what that trust should be based on.” Tom Findling, CEO of Conifers.ai, added that his concern with the Microsoft document is that it doesn’t answer the obvious question of how these models can possibly be controlled. “The document says the model should never resist being shut down, should stay within its scope, and shouldn’t hide what it’s doing. Those are all the right goals,” he said. “But the harder question is what happens when a highly capable agent doesn’t behave the way you expect. What actually stops it? As these systems get more autonomous, the safety boundary can’t just be that the model was trained not to do something. You need controls outside the model that limit what it can access, what it can do, and how far it can go.” Cybersecurity learned this lesson a long time ago, he pointed out. “You don’t secure a system by assuming it will behave correctly,” he said. “You assume something will eventually fail, get compromised, or act in an unexpected way, and you design the controls around that. AI needs the same mindset.” Frank Dickson, principal analyst at Dickson Research, contrasted Microsoft’s promise with Anthropic’s commitment, and found Microsoft lacking. “Microsoft’s document is shy on mechanism,” he said. “Compare the two on specifics. [Anthropic CEO] Amodei’s proposal names a third party, METR, and describes what access actually means: office badges, company laptops, employee-level visibility, and publishing rights Anthropic doesn’t get to edit. You can check whether that happened.” On the other hand, he noted, “Microsoft’s document says models should ‘fail tasks rather than violate the code’s rules,’ which is a real design principle, credit where it’s due, but there’s no named auditor, no verification method, and no stated consequence for a violation. Thirty-seven pages and it still won’t commit to a single verifiable check.” Dickson stressed that as long as agents routinely break their own rules, these vague promises won’t help. “Every frontier lab still gets jailbroken, still has agents that go off-script, still hasn’t closed the gap between what a model is instructed to do and what it can be induced to do,” Dickson said. “A values statement that skips the verification question isn’t a constraint, it’s a hope wearing a policy document’s clothes.”

When it comes to building robotic brains, many companies in the industry focus on putting intelligence directly into their robots. General Robotics is taking an alternate approach. The Redmond, WA company’s GRID system hosts its robotics intelligence layer in the cloud, allowing it to be delivered and deployed to robots operating remotely. GRID hosts the models, simulations, robot representations, data and agents through which robotic skills are developed, tested, orchestrated and deployed to robots on site. “Having one layer that looks after 50 to 100 robots is way more important than 50 or 100… robots that operate completely independently,” said General Robotics CEO Ashish Kapoor. While videos of sprinting robots crashing into pads — or humanoid robots kickboxing and breakdancing – often go viral, General Robotics focuses instead on industrial models. Its GPU-based GRID cloud provides a better base to ingest data and develop skills for specific verticals. “Our ability to onboard new robots, our ability to take them to deployment with new skills, our ability to connect very complex simulations to the reality is enabled through GRID,” Kapoor said. The company declined to talk in detail about its customers, but Kapoor offered an example of how the GRID-based system works. In a biolab, for instance, robots pouring liquid from beakers required in their modeling fluid tactile interaction and transparent materials. “There exists no good way of modeling those kinds of behaviors in software right now because [of]…all of those things,” Kapoor said. “…This is where agents came in.” AI agents were used to create a “hybrid simulator” that brought the pieces together in the cloud. The robots were then tested against the simulated model. “The physics of the robot was coming from one simulator and… the calculation of fluid dynamics was actually created by [the] agent itself by looking at the literature,” Kapoor said. What “otherwise would have taken us at least a year to build…was done in minutes. “GRID allows you to auto engineer those robots, retarget those skills, [and] create appropriate digital representations on its own,” Kapoor said. Humans are involved in safety checks to make sure robots behave according to specifications. While humanoid robots are the ones often drawing media attention, “the most value in robots in 18 months to 36 months is going to be realized in heavy industries,” he said. “These are ports, these are your energy infrastructures.” Although the company’s cloud-based GRID architecture has advantages, the system also faces some challenges — namely, lagging cloud-to-robot communication. “The round-trip time, the latency is of paramount importance,” Kapoor said. “The robot needs to have two loops.” He described that second loop as an “inner control loop…operating at a millisecond level…that’s on the edge.” Another issue involves companies that must meet strict regulatory and security requirements. For example, car companies are generally against putting proprietary information —whether that be designs for new fuel injection systems or new battery designs — in the cloud. Still, General Robotics remains optimistic its system can handle these kinds of issues long term. “Between commercial cloud deployment, between on-prem server deployment and…edge deployment, we do have a way of addressing many of these issues,” Kapoor said. In addition to continually improving its stack, the company also seeks out partners that can help it resolve challenges, Kapoor said. That robotics is a growing field seems obvious. Morgan Stanley in a 2025 study predicted the humanoid robot population would touch 1 billion by 2050 and be valued at $5 trillion. And a more recent study from Counterpoint Research estimated that between 50,000 and 85,000 humanoid robots would ship this year alone.

Microsoft is changing the destination address of two of its most popular services: starting this month, it will redirect M365 and Teams web users to copilot.cloud.microsoft and teams.cloud.microsoft, respectively. The Teams move is already under way, and Microsoft has now added M365 to the mix. The company announced the changes in two MessageCenter posts: MC1465764 (mirror) and MC1462915 (mirror). Organizations using these products are advised to update their systems and documentation to ensure continued access. Microsoft has told customers to review configurations on client devices, proxies, firewalls, secure web gateways, or other enterprise network controls to confirm that users can connect to the new addresses. Companies having trouble connecting should ensure that their environment aligns with the recommended network requirements for Microsoft 365 Copilot. Enterprises that had been blocking the new Copilot address to prevent employees accessing their personal Microsoft accounts can use Microsoft’s TenantRestrictions control to achieve their goals instead, it said. All redirects should be completed by early October, Microsoft said, advising companies that can’t meet the deadline to contact their account representative for help. Limited exceptions to the Teams redirect are possible until Dec. 31, 2026, but after that no further delays will be possible, it said.