AI

Scientists have been working on ways to decode brain activity for decades. Teams have developed brain implants that can be used to control robotic limbs and computers, and even to “speak.” The people who volunteer for trials of those devices play a vital role in that research, and they ought to be recognized for their contributions, says Ian Burkhart, who had one implanted in his own brain back in 2014. He’s been working on a set of guidelines for researchers developing brain-computer interfaces (BCIs). Ethics guidelines for clinical trials already exist. But BCI trials are different, says Jennifer Chandler, a professor of law and neuroethics at the University of Ottawa who also worked on the guidelines. “These are people who have a device embedded in their brain, are in trials sometimes for years and years … [and contribute] a significant investment of time, of the body, and of intellectual effort.” Burkhart was left paralyzed from the chest down following a diving accident in 2010, when he was 19 years old. Around four years later, he volunteered to become the first person to trial a brain implant designed for hand control. He had an array of 96 electrodes inserted into the part of his brain that controls movement. The electrodes were connected to a computer via a cable that essentially plugged into his head. Algorithms on the computer were designed to interpret those signals and deliver them to another sleeve of electrodes worn on Burkhart’s arm, allowing him to control it. The first movements were small, but Burkhart was ultimately able to use the system to pour beads from bottles, move small and delicate objects, and even play Guitar Hero, if only during lab visits. It was a scientific breakthrough. Burkhart took part in two years-long clinical trials over the seven-and-a-half-year period he had the electrode array implanted in his brain. “It was a fantastic experience,” he says. “But there’s certainly some things that, knowing what I know now, I would want to do a little differently.” In 2022, Burkhart cofounded BCI Pioneers Coalition—an organization through which people who have taken part in BCI trials can share their experiences, both with each other and with researchers. Over the last year, members of the group have been developing a set of guidelines based on those experiences. The idea is to help academics, clinicians, and companies run ethical trials that respect participants’ autonomy. The guidelines, published on Tuesday, lay out a list of recommendations that fall within 13 categories. One, for example, focuses on recognizing and compensating trial participants. During the trials that Burkhart participated in, “I was going into the lab two or three days a week, for four hours at a time,” he recalls. “That’s [easily] a part-time job if you just count the hours you’re putting in.” As the trial progressed, Burkhart says, he learned more and was able to give more constructive feedback on what was working and what wasn’t. He says he was a key member of the research team. Others who find themselves in a similar position might appreciate being listed as a coauthor on scientific publications or publicly named as a research collaborator, he says. Burkhart and his colleagues acknowledge that not everyone will want this, however.   The guidelines also address the thorny question of privacy. That’s important: The public will always be interested in cutting-edge biotechnologies like these—and the guidelines say researchers also have a duty to prepare participants for the possibility of media attention. In some cases, it might be difficult to completely conceal the identity of trial participants. Things get even more complicated when it comes to the privacy of participants’ brain data. Research teams will need to access, analyze, and—at least temporarily—store this data. Burkhart wasn’t too bothered about that in his case, where scientists were collecting neural activity related to movement. But people who have brain implants designed for turning thoughts into words might feel differently.  The guidelines state that brain data “should be understood as the participant’s personal information.” They are pushing for participants to be able to access their own brain data whenever they want—something that has not always been possible—and want to make sure people aren’t prevented from sharing it with their own doctors because of embargoes or nondisclosure agreements (NDAs) designed to keep findings under wraps from competitors. “It’s not super common for there to be firm NDAs within a lot of the trials,” says Burkhart. “But it’s something where we want to make sure that that does not happen.” The guidelines also address what comes after a trial. Around five years after Burkhart had his device implanted, the research team ran into some issues with funding. When we last spoke a few years ago, he recalled how he was told he would have to have the implant removed and could have it put back in once the team had secured funding. (And we’re talking about brain surgery here.) Ultimately, he kept the implant until 2021, when it was removed after an infection started to develop at the point where the cable led into his scalp. But losing the device after all those years was difficult; he’d regained the ability to move his arm—and then lost it again. Rita Leggett, an Australian woman who had a device implanted in her brain to treat her epilepsy, has said she felt she “became one” with her implant. When the company that made it went bust two years later, the device was removed against her will—an act ethicists have described as a potential breach of human rights. Researchers should instead commit to maintaining brain devices for a “reasonable period” after a trial ends, according to the guidelines. Participants should be given plenty of time to prepare for the surgery to remove it. And researchers should set aside extra funds to support trial participants who have to do that. Some of the guidelines are vague and may overlap with existing ethics standards. They don’t all spell out explicit solutions. They

The self-styled visionaries of Silicon Valley, leaders like OpenAI chief executive Sam Altman and SpaceX boss Elon Musk, have built careers claiming to see the future. Long-term vision is currency: Meta’s Mark Zuckerberg built a 10-year road map for Facebook in 2016, and Google famously told investors before its public offering that it was unconcerned with short-term results. And yet there they all were, a mere five weeks before the midterm elections, gathered at the White House to smile for photos with President Trump and sign on to his “Super Intelligence” agenda. Even by the diminished standards of today’s corporate governance, this was a shockingly foolish decision. Despite dozens of polls showing a rising tide of anti-Trump sentiment and bipartisan rejection of AI and data centers, the futurists of the tech elite cast aside long-term corporate interests for short-term political favor. And now, their industry is very squarely in the new victors’ crosshairs. For an industry obsessed with pattern recognition, their decision to lash themselves to President Trump has proved to be a catastrophic misreading of the political landscape. Now, not only have they jeopardized their companies’ fates, but they’ve also raised questions about their vision and judgment.  What were they thinking? The people who work for them are just as confused as the rest of us. Tech workers I’ve spoken to at OpenAI, Meta, and other firms have noted an absence of commentary from their leaders post-election. Zuckerberg has yet to address the results internally, while an employee at another AI firm said that “leadership has been pretty silent on Slack.” Most employees share concern about potential anti-data-center legislation and are curious how leadership will adjust to a Democratic Washington, DC. But more than anything, they don’t understand how their bosses didn’t see what was so blindingly obvious to everyone else. In Silicon Valley, the Gretzkyism “skate to where the puck is going, not where it has been” is a management maxim; it preaches anticipation and foresight over reactiveness to the present moment. Tech entrepreneurs love a good “pivot,” and companies that can’t change with the times get disrupted and die. Which is why I find it so strange that Zuckerberg, OpenAI cofounder Greg Brockman, and so many of Silicon Valley’s other prominent AI leaders and venture capitalists have put themselves in a potentially inescapable political bind. In going all in on President Trump’s cultural and business agenda instead of playing both sides of the aisle as corporate America usually does, they’ve built identities they may not be able to pivot out of. Will anyone really believe that Palantir CEO Alex Karp is suddenly progressive after a year of promising “violence and death to our enemies” and telling investors that the company is “completely anti-woke”? I believe this is less a surprise than a very public exposure of a fundamental reality: that tech’s most visible leaders are not all-knowing visionary masterminds after all. Insights about social media or artificial intelligence do not make Zuckerberg or Altman experts on politics, the economy, or much else outside of technology. The instant gratification of presidential access, subsidies, and a say in their own regulation blinded them to the basics of political math that they clearly did not understand. There’s a reason the chief executives of Chase, Boeing, and other Fortune 500 staples don’t go on podcasts to talk about “masculine energy,” or call the president from the stage of a MAGA-friendly AI summit to label fears about AI a “hoax” and a “scam.” This is the learned wisdom of decades operating in an environment where the customer changes every four years but contracts run for 10. Most companies play to both sides of the aisle because they know that their business models and multibillion-dollar projects must outlast the current administration. Instead, Big Tech tied itself to Trump and made its most important goals and investments—many related to grid access and data center permits—immediate targets for Democrats. In tech, you can always sunset a feature or update some code. But in politics, there is no undo button for failures that change public perception. The tech elites’ decision to visibly align their businesses with a temporary movement has left them no credible way to pivot now that the climate has changed. As one Meta employee who has worked closely with Zuckerberg put it, “Mark is a long-term thinker, but he never charts a plan of how to get there. It’s spray and pray and see what happens.” As for whether employees expect the company to do a strategic postmortem? “Facebook has never really done that.” “Move fast and break things” might work in a startup, but it’s a terrible strategy for building reputational goodwill. Previous technological leaps, like the rollout of fiber-optic technology and wireless infrastructure, avoided partisan politics. But by kowtowing to the Trump administration, Silicon Valley made artificial intelligence a culture-war issue. Now, when citizens vote on data center proposals from California to Florida, they are no longer considering only technical or economic concerns, but political ones. Tech leaders didn’t see this coming because they were too focused on the urgency of the AI race to understand how reputational damage could undermine the intangible goodwill—and ultimately the very tangible infrastructure—necessary to win it. The industry’s misreading of the moment has already punctured one myth Silicon Valley holds dear: that its leaders can always see what’s next. Aaron Zamost is a tech communications consultant and a former executive at Square.

Representatives from Nvidia told Illinois representative Bill Foster that it would be technically possible to implement the anti-smuggling measures called for in the Chip Security Act, Foster told MIT Technology Review in an interview on Thursday, even while an industry group claimed they were “potentially infeasible.”  Foster, a Democrat who serves on the Committee on Science, Space, and Technology, is a cosponsor of the act, which would introduce geotracking requirements to verify the location of advanced AI chips in order to prevent their smuggling to China. It is one of three chip-security bills under consideration for inclusion in the Senate’s National Defense Authorization Act, each of which is drawing renewed attention after Nvidia CEO Jensen Huang’s resignation and reported affair with an alleged Chinese spy. When the Chip Security Act was introduced, in May 2025, lawmakers said the technology to implement it existed, something Nvidia has not publicly confirmed. In March 2026, the Semiconductor Industry Association, of which Nvidia is a member, explicitly opposed the bill and wrote that its mandates for chip tracking were “new, untested, and potentially infeasible.”  But privately, Nvidia representatives working with Foster told the lawmaker that the company could implement them.  “They never put anything in writing,” Foster says, but during technical discussions on the possibility of such safeguards, their reaction was “yeah, of course we can do this.” This private acknowledgment to Foster goes further than anything Nvidia has previously made public.  Other technology Nvidia has built for its customers suggests that the company can indeed offer accurate tracking of its chips’ locations. In December 2025, for example, it announced a new tool called Fleet Intelligence that would allow customers to monitor how its advanced chips were performing across different data center locations. In doing so, the tool could indicate which country chips were in, Nvidia confirmed to Reuters, but the company would not elaborate on whether this technology could be used to prevent smuggling.  Nvidia did not respond to requests for comment or confirmation of Foster’s characterization. In early October, when lawmakers were reportedly increasingly concerned about Nvidia’s role in smuggling, MIT Technology Review asked the company if Fleet Intelligence could be used to prevent the practice. It declined to answer directly, instead sending a statement: “Governments and media have found insignificant diversion of controlled US products—a drop in the bucket compared to the ocean of domestic compute China already has.” The recent fallout at Nvidia involves a number of people and companies alleged to be involved in chip smuggling. Alice Huang, with whom Jensen Huang reportedly had an affair, runs Megaspeed, a Singaporean firm now under investigation for smuggling advanced GPUs to China. A US Commerce Department inquiry into the company was reportedly halted after invention by Nvidia. Megaspeed reportedly bought chips from Supermicro, a company that packages Nvidia’s chips onto servers.  The Chip Security Act as drafted would require a mechanism to track the locations of chips, either by inspections of facilities, attestations by companies, or ping-based location mechanisms. The ping-based technique, the hardest option to circumvent, would require the GPU to send a signal of its location when it comes online. In September, the Center for Security and Emerging Technology found it would be the most cost-effective option. Foster said he did not ask Nvidia’s representatives why they weren’t voluntarily adopting such chip-tracking tools if they could technically do so. But if the requirements of the bill pass as part of the Senate’s National Defense Authorization Act, the new tracking mechanisms would only tell investigators where the chips are, not offer a way to disconnect them if it’s revealed they have been smuggled. Foster told MIT Technology Review he intends to introduce new, stricter legislation in the new year that would offer a workaround. The new rule would require licenses to operate advanced GPUs, and before a GPU could connect, its valid license would need to be verified.   In the meantime, the controversy at Nvidia, Foster says, has fueled support among his colleagues for the rules in the bill: “There will be a lot of political pressure, in the wake of everything that’s happened, on the companies to actually do a very good job of the geographic validation of where the chip actually is.”

arXiv:2611.13477v1 Announce Type: cross Abstract: Integrating human feedback into language models has gained increasing attention as a means to align model outputs with human preferences, with Reinforcement Learning from Human Feedback (RLHF) serving as a prominent framework for preference-based alignment. Recent studies have incorporated eye-tracking (ET) data as an additional supervisory signal for reward model training, grounding reward learning in human reading behavior. Building on this approach, we extend prior fixation-centric models by leveraging established findings from cognitive psychology on human letter recognition. Specifically, we model visual attention as a spatially graded distribution centered on the current gaze location and incorporate this representation into reward model training. Our results show that the reward model trained with the proposed gaze distribution achieves higher preference prediction accuracy than a baseline model. Moreover, some results from our experiments are qualitatively consistent with established findings on human attention during reading, suggesting that they potentially capture cognitively meaningful aspects of human attention allocation.