Tech News
-
House House reveals generative AI used for under 0.1% of Big Walk code
The boundary between technical assistance and the use of generative AI has become a critical point of contention for indie developers, as evidenced by a recent disclosure from House House. The studio, known for the hit Untitled Goose Game, has revealed that Large Language Models (LLMs) were utilized during the final stretch of development for their latest title, Big Walk.
This admission comes as a voluntary disclosure tied to the game's submission for the Independent Games Festival (IGF) awards. The timing is particularly sensitive, as a group of IGF jurors—including Sam Barlow, Austin Walker, and Lena Reine—recently signed a statement declaring their intention to decline votes for any game that employs generative AI in any capacity. The jurors specifically noted that AI-generated code is not an exception to these concerns, placing Big Walk under scrutiny as it was being considered for the awards.
Technical Implementation and Scope
To clarify the extent of the technology's role, House House stated that LLMs were used sparingly and exclusively to resolve difficult technical bugs and challenges. The developers estimate that well under 0.1% of the code in Big Walk was influenced by this usage, and they emphasized that generative AI was not used in any other context during the production of the game.
The studio provided a specific example of this usage involving the Unity Universal Render Pipeline (URP). The developers encountered an issue where shadow casters behind the player were not being culled, causing the game to run 50% slower for approximately 20 seconds during sunrise and sunset. After extensive internal research and a consultation with a Unity engineer who concluded the fix was impossible, the team turned to an LLM chatbot. While the chatbot initially provided several dead ends, it eventually suggested a fix that the developers had to modify extensively. The final result was a working solution consisting of roughly 10 lines of code.
Policy and Industry Standards
Despite the minimal amount of code involved, House House described the lack of an internal AI usage policy during development as a failure. The studio expressed concern that voluntary disclosures of this nature could contribute to the normalization of generative AI or foster a perception that the technology is inevitable. In their statement, the developers apologized if their usage did not meet the standards of those working to limit the harm generative AI may cause to the gaming industry and the wider world.
This disclosure was not made on Steam, as the platform currently does not request the disclosure of chatbot usage or AI-generated code. House House noted that they are only speaking publicly on the matter now because it has become a matter of public interest. This transparency follows the commercial success of Big Walk, which launched in August 2026 to critical acclaim and reached one million sales in less than a week.
The situation highlights a growing rift in the industry regarding the definition of "generative AI." While some view LLM-assisted coding as a pragmatic tool for debugging, others, including the aforementioned IGF jurors, view any such integration as a breach of creative integrity. As House House works to implement a more comprehensive internal policy, the outcome of the IGF deliberations may serve as a bellwether for how the independent community treats AI-assisted technical fixes moving forward.
-
Nvidia CEO Jensen Huang faces national security crisis over alleged affair with Chinese spy
The admission of an extramarital affair by Nvidia CEO Jensen Huang has evolved from a personal scandal into a national security crisis, as U.S. intelligence alleges the other party was a Chinese spy. The situation has placed the semiconductor giant at the center of a geopolitical firestorm, raising questions about whether the world's most valuable AI chipmaker was the target of a sophisticated "honeypot" operation designed to help Beijing achieve AI dominance.
The woman at the center of the controversy, Alice Huang—identified in travel records as Chinese citizen Huang Le—is alleged by U.S. intelligence to have been recruited by China's Ministry of State Security. While China has dismissed these claims as a smear, U.S. officials believe she used her background as a journalist and finance executive to embed herself in high-level AI circles. This relationship was not merely social; Alice Huang served as a director at Megaspeed International, a Singapore-based cloud computing and data center provider that has become one of Nvidia's largest customers in Asia, purchasing an estimated $4 billion in semiconductors.
The Megaspeed Connection
The link between the CEO and Megaspeed has triggered intense scrutiny over the diversion of restricted AI hardware. Megaspeed is accused of smuggling restricted Nvidia chips to Chinese companies, a claim bolstered by reports that the firm imported 50, companies reported 50,000 more chips than were actually tallied in its data centers. While Megaspeed denies any wrongdoing, the timing of these revelations is particularly damaging given Nvidia's struggle to navigate U.S. export controls. Since 2022, Nvidia has attempted to maintain its footprint in China by re-engineering products like the H20 to comply with regulations, though the company eventually revealed $5.5 billion in costs from canceled orders for that specific chip.
This tension is further complicated by a discrepancy in reported revenue. While Nvidia recently stated that China and Hong Kong accounted for roughly 8% to 9% of its revenue, short seller Culper Research alleged the exposure is much higher, claiming over 20% of fiscal 2026 compute revenue was driven by China through illegal GPU diversion and Southeast Asian intermediaries. The possibility that restricted Blackwell chips reached China via Megaspeed is now a primary focus for investigators.
Legal and Political Fallout
The political ramifications are equally severe, as Jensen Huang has maintained a close relationship with President Donald Trump, advising the White House on AI safety and pushing for looser export restrictions. This influence bore fruit in December when the administration allowed Nvidia to ship H200 chips to China in exchange for a 25% cut of sales. However, the revelation of the affair has weakened Huang's standing in Washington. President Trump confirmed that the FBI is investigating the matter, though he noted that Huang claimed to have no knowledge that Alice Huang was a spy.
The corporate fallout was immediate, with Nvidia's stock dropping approximately 4.27% following the CEO's admission, erasing hundreds of billions of dollars in market value. This decline prompted a proposed class action lawsuit filed by shareholder Mark Kistenmacher, which accuses Nvidia and Huang of misleading investors by describing dealings with Megaspeed as "normal and not unusual." Additionally, three House committees have requested closed-door briefings, and an independent investigation has been launched to determine if state or corporate secrets were compromised.
Further complicating the legal landscape is a report that a U.S. Commerce Department probe into Megaspeed was abruptly transferred to the Justice Department after Nvidia's general counsel contacted the Commerce Department seeking information about the investigation. While Nvidia maintains that no secrets were stolen, the intersection of a personal affair, alleged espionage, and suspected chip smuggling has left the company vulnerable.
The broader implications suggest a tightening of the leash on AI hardware exports. With the political mood in Washington shifting following recent midterm elections, there is growing momentum for legislation that would grant Congress direct oversight of chip export licenses. Future regulations may include requirements for chipmakers to install security mechanisms to track the physical location of hardware, ensuring that advanced AI capabilities do not reach restricted territories through third-party intermediaries.
-
AMD confirms GPU driver bug that causes excessive wear and reduces SSD lifespan
A critical bug in AMD's GPU drivers is reportedly causing some systems to continuously write data to the system drive, potentially accelerating the wear and tear of SSDs. The issue centers on the AMD External Events Utility (EEU), a component installed as part of the AMD Software: Adrenalin Edition, which in some configurations enters a loop of constant data writing even when the computer is idle.
The scale of the data being written is significant, with affected users reporting write speeds ranging from 0.3 to 5 MB/s. At the upper end of that spectrum, a system could generate approximately 432 GB of host writes per day, which totals nearly 13 TB over a full month of uninterrupted activity. Some Radeon owners have claimed this has had a tangible impact on hardware longevity; one user reported that their SSD's health indicator declined by roughly one percent every month for 16 months, though this specific claim remains unverified.
While these numbers are alarming, it is important to distinguish between reported host writes and actual physical NAND flash degradation. SSD controllers employ caching and wear-leveling techniques to manage incoming data, meaning the reported activity does not always translate directly to a linear reduction in drive lifespan. Nevertheless, AMD has acknowledged the problem, with company representatives confirming as of October 21 that they are investigating the cause. The company has attributed the behavior to an apparent interaction or loop between the External Events Utility and a PCIe-related component.
Detection and Scope
The bug appears to affect both Windows 10 and Windows 11 systems across various driver versions, including Adrenalin 26.19.2 and older releases. Interestingly, the issue was not mentioned in the release notes for Adrenalin 26.20.2, which launched on October 30. Because the problem does not seem to follow a clear pattern across specific hardware combinations or driver versions, users are encouraged to manually verify if their systems are affected.
To check for the bug, users can open the Windows Resource Monitor by typing "resmon" into the Run command (Windows + R) and navigating to the Disk tab. The signal of the bug is sustained, continuous write activity from the processes atieclxx.exe or atiesrxx.exe while the PC is otherwise idle. A brief spike in activity is typical for background tasks, but the bug is characterized by a persistent flood of data that does not let up.
Mitigation and Resolution
Until a formal patch arrives, there are a few unofficial workarounds, though each carries a performance or functional penalty. One method involves using the Windows Services application (services.msc) to stop or disable the AMD External Events Utility. While this halts the excessive writes, it may break Radeon hotkeys, display detection, and FreeSync features. Alternatively, some users have found that disabling Resizable BAR (ReBAR) stops the activity, though this is generally undesirable as ReBAR is designed to improve gaming performance by optimizing GPU memory access.
AMD has stated that a resolution is expected to be delivered in driver version 26.11.1, although a confirmed release date for this update has not yet been provided.
The situation highlights the risks associated with background utilities that lack sufficient safeguards against infinite write loops. For now, the safest course of action for Radeon owners is to monitor their disk activity via Resource Monitor and avoid unofficial forum patches until the 26.11.1 driver is officially released and verified.
-
Anthropic releases Claude Fable 5.2 and Mythos 5.2 models amid reported IPO plans
Anthropic has released Claude Fable 5.2 and Mythos 5.2, targeting the most complex software engineering and research tasks with a significant reduction in API costs. These new frontier models are designed for ambitious, open-ended technical work, marking a shift toward more autonomous AI agents capable of handling high-value software projects and scientific research.
While Fable 5.2 and Mythos 5.2 share the same underlying architecture, they are distinguished by their safety frameworks and availability. Fable 5.2 is broadly available to the public, including through paid Claude plans and major cloud providers like AWS, Google Cloud, and Microsoft Azure. In contrast, direct access to Mythos 5.2 is restricted to vetted cybersecurity and life sciences organizations through Anthropic's Project Glasswing and verification programs, though Claude Enterprise customers can access its vulnerability detection via Claude Security.
Performance and Economics
The release is accompanied by a 20% reduction in API pricing compared to the previous 5.1 generation, alongside a 30% increase in output generation speed. For users requiring even more velocity, an optional fast mode is available that can speed up token generation by up to 2.5 times. The updated pricing structure is as follows:
- Standard: $8 per million input tokens / $40 per million output tokens
- Fast Mode: $16 per million input tokens / $80 per million output tokens
These economic shifts are paired with substantial coding gains, particularly when utilizing different "effort" settings. On the FrontierCode 2.0 benchmark, which measures whether AI-generated changes would be accepted by open-source maintainers, Fable 5.2 scored 49.1% at its "Xhigh" effort setting, a significant jump from the 32.4% achieved by Fable 5.1. Interestingly, Fable 5.2 at a "High" effort setting achieved a score comparable to Opus 5.5 at its highest setting, but at roughly half the cost per task. This makes Fable an attractive option for demanding assignments like codebase migrations or stubborn bug diagnosis, though Anthropic continues to recommend Opus 5.5 as the primary "daily driver" for most general coding and knowledge work.
The model's capacity for complex, agentic orchestration was demonstrated in an internal experiment where Fable 5.2 coordinated over 1,000 subagents over eight days to build a prototype operating system capable of booting on a laptop. This level of coordination suggests a move toward AI acting as an "incident commander" that can delegate information gathering to faster models while managing the overall project architecture.
Security and Compliance
The Mythos 5.2 variant specifically pushes the boundaries of cybersecurity engineering. In an internal CVE Rediscovery evaluation involving 1,829 previously disclosed vulnerabilities, Mythos 5.2 identified 89.6% of flaws, up from 78.3% in the previous version. While this improves the capabilities of defenders in controlled enterprise settings, some analysts warn that such advances in vulnerability discovery could similarly empower attackers, accelerating an AI-powered arms race.
Despite these gains, operational hurdles remain for enterprise deployment. Reliability without human supervision is still a primary concern, and the coordination of multiple agents across internal teams and enterprise tools remains difficult. Furthermore, data retention policies have emerged as a point of contention for regulated industries. While some customers are eligible for zero data retention, others face a 30-day retention period for safety monitoring. To address this, Anthropic has introduced Enterprise Frontier Safeguards in private beta, allowing automated safety monitoring to run directly within a customer's own infrastructure.
These developments signal a transition toward highly autonomous software development, where the bottleneck is shifting from raw model capability to the reliability of agentic workflows. As AI models become more economical and capable of managing multi-day projects, the focus for enterprises will likely move toward establishing rigorous human-in-the-loop approval processes and isolated environments to manage the risks of autonomous execution.
-
Rockstar Games confirms GTA 6 targets 30 FPS on consoles and PSSR on PS5 Pro
Rockstar Games has finally detailed the performance targets for Grand Theft Auto VI, confirming that the massive open-world title will target 30 FPS across all supported consoles. This decision prioritizes the density and visual fidelity of the state of Leonida over higher frame rates, even on the most powerful hardware available.
By targeting 30 FPS on the PlayStation 5, PlayStation 5 Pro, and Xbox Series X|S, Rockstar intends to push the current generation of hardware to its absolute limits. The developer stated that this approach is necessary to maintain the visual richness and inhabitant density of the game's environments. While the frame rate target remains consistent across the board, the PlayStation 5 Pro will distinguish itself through specific fidelity upgrades, including ray-traced shadows and the use of PlayStation Spectral Super Resolution (PSSR) for improved image clarity.
Technical Specifications and Speculation
The implementation of PSSR on the Pro hardware has already drawn interest from the technical community. According to datamining of the game's preload metadata, GTA VI may be utilizing a newer iteration of Sony's AI upscaling technology than other current Pro-enhanced titles.
- PSSR Version 13.60 (GTA VI preload)
- PSSR Version 13.00A (Cyberpunk 2077, Assassin's Creed Shadows)
This discovery has led to further speculation regarding the game's output. Metadata references to a 120Hz output mode have emerged, which some suggest could point toward a potential 40 FPS mode—a middle ground between 30 and 60 FPS that fits evenly into a 120Hz refresh cycle. However, this remains unverified, as Rockstar has not explicitly confirmed any frame rate options beyond the 30 FPS target. Additionally, the absence of QSSR references in the preload data suggests the base PS5 may not receive the same upscaling improvements as the Pro model, though this is not yet conclusive.
Platform Integration and Accessibility
Beyond the raw performance targets, Rockstar has emphasized deep integration with the PlayStation 5 ecosystem. The game features high-fidelity haptic feedback for the DualSense controller, which players can adjust via accessibility settings, and utilizes Tempest 3D Audio for improved spatial sound rendering. Combined with optimized load times, Rockstar asserts that these features make the PlayStation 5 consoles the best place to experience the title.
Xbox players will have a different set of options, most notably the inclusion of Xbox Cloud Gaming. This allows the game to be streamed to Xbox Series X|S and Xbox One consoles, though the cloud-streamed version will utilize the graphical settings of the Xbox Series S. While the specific performance of the local Xbox Series S version has not been detailed, it is expected to be the most limited version due to the console's hardware specifications. Both PlayStation and Xbox users will also have access to Remote Play functionality.
As the November 19 launch date approaches, the logistical scale of the project is becoming apparent. Pre-loading is already active, requiring players to clear 212 GB of storage space. This massive footprint aligns with the ambition described by Take-Two Interactive, which recently noted that pre-sales for the title are running far ahead of company expectations.
These performance decisions highlight a broader trend in high-fidelity open-world design, where developers are forced to choose between fluid motion and extreme environmental detail. By locking the target at 30 FPS, Rockstar is betting that players will value the "absolute limits" of visual density over the responsiveness of a higher frame rate. The industry will now be watching to see if the PS5 Pro's PSSR and ray-tracing enhancements are enough to offset the lack of a 60 FPS mode for the most anticipated release in gaming history.
-
Sony announces Japan-exclusive PlayStation Portal Player handheld featuring internal storage for future functionality
Sony is edging closer to a standalone handheld experience with the reveal of a new, storage-equipped PlayStation Portal. While the device has functioned strictly as a streaming accessory since launch, the introduction of internal storage suggests a shift in how Sony views the future of portable PlayStation gaming.
The company has officially unveiled the PlayStation Portal Player, a Japan-exclusive model arriving in Midnight Black on December 10, 2026. Priced at 39,980 yen—roughly $252 to $253—the device is strictly region-locked, supporting only the Japanese language and requiring a PlayStation account registered in Japan. Pre-orders for the unit are scheduled to begin on November 13.
Hardware and Future Functionality
The most significant departure from the global model is the addition of internal storage. Sony has not yet disclosed the exact capacity of this storage or the specific file types it will support, but senior product manager Takuro Fushimi noted that the addition was made with an eye toward "future functionality." While this has sparked speculation regarding the possibility of native game execution or offline capabilities, Sony has not confirmed that the device will support downloadable PS5 titles.
Beyond the storage, the hardware remains largely identical to the existing Portal. It retains the 8-inch 1080p LCD screen with 60fps support, 3D Audio, and the haptic feedback and adaptive triggers found in the DualSense controller. The only other notable change is the branding; Sony has dropped the word "Remote" from the product name, rebranding it simply as the PlayStation Portal Player. This change follows the 2025 addition of cloud streaming capabilities, which expanded the device's utility beyond its original role as a Remote Play tool.
Streaming and Market Strategy
Despite the new storage, the PlayStation Portal Player continues to rely on two primary methods of gameplay. Users can utilize Remote Play to stream games from a paired PS5 console over Wi-Fi, or they can use cloud streaming via a PlayStation Plus Premium subscription to play supported titles directly from Sony's servers. To facilitate this, Sony is including a one-month subscription to PlayStation Plus Premium with the new model. For a stable experience, Sony recommends a connection speed of 15Mbps, though a minimum of 5Mbps is required.
This Japan-only release appears to be a strategic move to maintain competitiveness in Sony's home market, where the success of the Nintendo Switch and the high cost of PlayStation hardware have pressured the brand. This mirrors a previous tactic where Sony released a region-locked, more affordable digital edition of the PlayStation 5 in Japan. By pricing the Portal Player identically to the global model despite the added storage, Sony is likely attempting to increase accessibility for domestic customers.
The move toward internal storage serves as a bridge between a pure streaming peripheral and a true gaming handheld. While the current Portal Player remains an incremental update, it lends weight to ongoing rumors regarding a more powerful PlayStation 6 handheld. Such a device is speculated to utilize an AMD-made chip and PlayStation Super Spectral Resolution (PSSR) upscaling to run PS5 games natively. Whether the new internal storage is a precursor to that next-generation hardware or a way to introduce local game saves and enhanced cloud features remains to be seen, but it marks the first time Sony has added permanent storage to the Portal line.
-
GTA 6 may arrive on PC sooner to combat AI-generated clones, former Rockstar directors claim
The battle for Grand Theft Auto 6 is moving beyond console exclusivity as generative AI threatens to bridge the gap between the PlayStation 5 and the PC. While the title is slated for a November 19, 2026, release on consoles, the industry is now grappling with the possibility of AI-driven unofficial ports that could undermine Rockstar Games' traditional release cadence.
Obbe Vermeij, a former technical director at Rockstar North, suggests that the pressure to accelerate the official PC port may stem from the rise of generative AI. Vermeij notes that tools capable of automating the porting process—creating what some call "slopcomps"—could allow unofficial versions to appear shortly after the console launch. While these AI-generated versions may lack the polish of an official Rockstar release, Vermeij warns they could still significantly impact sales by providing a playable, albeit inferior, alternative for PC users. He observes that while such a process would require tens of thousands of dollars in tokens and significant time, the massive demand for a title like GTA 6 makes it a likely target for those attempting AI-driven ports.
The Rise of AI Porting
This threat is not merely theoretical, as evidenced by an open-source project known as AnyPS5. This AI-powered project on GitHub attempts to convert PS5 game executables into programs that run natively on Windows and Linux by swapping Sony’s system libraries for its own versions rather than relying on emulation. According to project updates, 100% of PS5 GPU shaders have been ported, and system library support has reached between 86% and 88%. While some speculators believe this could allow GTA 6 to run on PC at launch with higher performance than the console version, current results remain rudimentary; demos have shown GTA 5 running at 27 FPS and other titles ranging from 4 to 60 FPS. Community members also note that mandatory firmware updates and DRM could create significant roadblocks for the project.
In response to these shifting technical landscapes, reports suggest Rockstar may be moving the official PC release closer to the console window than in previous cycles. While the studio has a history of delaying PC ports—GTA 5 took 18 months and Red Dead Redemption 2 took 13—industry sources indicate GTA 6 could arrive on PC in the first quarter of 2027. This potential acceleration aligns with comments from Take-Two Interactive CEO Strauss Zelnick, who recently noted during a shareholders meeting that the PC platform is becoming increasingly important to the company's overall strategy.
Beyond the threat of AI, several economic factors are pushing the PC port forward. With the PC gaming market projected to reach one billion players by 2028 and the rising cost of home consoles, the potential for revenue is immense. Analysts project launch week revenue for GTA 6 between $3 billion and $5 billion. A quicker transition to PC would allow Rockstar to capitalize on the lucrative microtransactions associated with a new iteration of GTA Online more efficiently, rather than relying on the "double-dipping" sales practice of staggered releases.
The emergence of AI-assisted reverse engineering marks a fundamental shift in how AAA titles are protected. Historically, the sheer scale and complexity of development acted as a natural barrier against rip-offs, but the speed of modern generative AI is eroding that advantage. Whether through official acceleration of the release date or a legal battle against "vibe coders," the launch of GTA 6 will likely serve as a litmus test for how publishers manage intellectual property in an era where the technical barrier to porting a game is rapidly disappearing.
-
Botched Nier: Automata PC update inflates game file size from 50GB to 366GB
A routine localization patch for NieR: Automata has inadvertently turned the action RPG into a storage nightmare, demanding a staggering 366GB of free space for installation. This massive inflation in file size has left Steam users bewildered, as a game that typically requires roughly 50GB of storage suddenly asks for more than seven times that amount before the download can even begin.
The issue stems from an update published on November 5, 2026, which was intended to be a straightforward addition of localization support for Korean, Simplified Chinese, and Traditional Chinese. This marked the first Steam update for the title since July 2021, but instead of a seamless transition, players found that Steam refused to initiate the installation unless 366GB of free space was available. This is despite the game's official store page still listing a requirement of only 50GB.
The DLC Duplication Glitch
While Square Enix has officially confirmed the issue and stated they are currently investigating the cause, player reports suggest a more specific technical failure. According to these reports, the update appears to be incorrectly assigning the full size of the base game to each of the eight DLC packages bundled with the Game of the YoRHa Edition. This effectively forces Steam to pre-check for nine separate copies of a 40GB game, artificially inflating the initial storage requirement.
It is important to note that this 366GB figure is a requirement for the installation process rather than the actual footprint of the game on the drive. Reports indicate that once the game is successfully installed, the file size reverts to the appropriate 40GB to 50GB range. However, for many users with limited SSD capacity, the initial requirement acts as a hard wall, preventing the game from being installed at all.
To mitigate the problem, Square Enix issued a temporary workaround on November 6. They recommend that players navigate to the "Properties" menu for NieR: Automata in Steam, select the "DLC" tab, and uncheck every listed DLC item. By doing so, Steam will only download the base game at its intended size, allowing players to bypass the inflated storage check.
The bug has already triggered a wave of negative Steam reviews from frustrated users who found themselves unable to launch the game without clearing massive amounts of data from their drives. The absurdity of the requirement is highlighted by the fact that the 366GB figure exceeds the announced installation size of other massive modern titles, such as the 212GB requirement for GTA 6.
This anomaly serves as a stark reminder of the growing tension between game installation sizes and available hardware storage. While this specific instance was a bug, the industry has seen a steady climb in storage demands, with titles like Call of Duty: Modern Warfare and Black Ops Cold War already surpassing 200GB. As developers continue to push for higher fidelity, the risk of "SSD-choking" requirements—whether accidental or intentional—will likely become a more frequent point of contention for PC gamers.
-
OpenAI rebrands data centers as compute campuses amid rising protests and infrastructure concerns
The rapid expansion of AI infrastructure has hit a wall of public resistance, transforming the physical sites of computation into political lightning rods. As the massive power and water requirements of generative AI strain local grids and environments, the industry is facing a growing "culture war" that pits tech giants against the communities hosting their hardware.
In an effort to pivot away from the negative connotations associated with AI data centers, OpenAI is proposing a shift in terminology, rebranding these facilities as "compute campuses." According to a company spokesperson, the term "compute campuses" more accurately reflects the scale and long-term nature of the investments in energy infrastructure and skilled labor. This move comes as public sentiment sours; a recent Pew Research poll indicated that 54% of Americans view AI data centers as detrimental to the environment, a sentiment that has become a prominent issue in the U.S. midterm elections.
This shift in terminology is accompanied by a broader strategic framework designed to ameliorate local tensions. OpenAI has introduced a National Community Compact and a Compute Campus Policy Blueprint, which together outline how the company intends to integrate its infrastructure into local ecosystems. The National Community Compact establishes a set of mandatory commitments for any compute campus owned and operated by OpenAI, ensuring that residents are engaged early in the planning process and provided with a dedicated local point of contact.
Community Commitments
To move beyond mere optics, OpenAI has pledged specific operational standards to ensure that the burden of these facilities does not fall on local taxpayers or the environment. These commitments include:
- Full payment for all electricity and required infrastructure costs, including financial assurances to protect other customers if a project is canceled.
- The implementation of site-specific water plans that prioritize low-water or closed-loop cooling.
- The publication of independent noise studies and construction plans before and during operations.
- A prohibition on the use of non-disclosure agreements (NDAs) with public officials, except for narrow security or proprietary secrets.
The company is already attempting to demonstrate the tangible benefits of this approach through targeted local investments. In Effingham County, Georgia, this resulted in a $4.8 million grant to provide free school meals for 14,000 students, while in Pike County, Ohio, OpenAI and SB Energy committed $6 million toward childcare services. These initiatives are intended to show that the economic opportunities created by AI infrastructure can be shared with the residents of the hosting communities.
Industry-Wide Policy Proposals
While the Compact governs OpenAI's own sites, the Compute Campus Policy Blueprint is a proposal for federal, state, and local policymakers to establish industry-wide rules. This blueprint suggests that state utility commissions should require all large electricity customers to cover the additional costs they impose on the grid. It also advocates for the acceleration of transmission upgrades by Congress and the expansion of tax incentives for clean electricity.
Central to this policy is the "Grid Protectors Program," a proposal to make power flexibility a core element of campus design. This program would allow OpenAI to reduce its electricity consumption during periods of peak demand, ensuring that nearby residential heating and cooling systems remain functional. This approach stands in contrast to other industry players; for instance, Google has reportedly proposed paying residents to turn off their own appliances during high-demand periods so that its data centers can continue to operate.
These efforts to rebrand and regulate arrive at a critical juncture of global instability regarding AI infrastructure. In Europe, coordinated protests are planned across 15 countries, with activists demanding a moratorium on new large-scale developments. These campaigners cite the "AI goldrush" as a grab for scarce land and water, pointing to incidents such as a Google subsidiary reportedly felling 250,000 trees illegally in Finland to make room for a server hall.
The success of the "compute campus" rebrand will likely depend on whether these policy shifts can outpace the growing political momentum against "dirty datacenters." As governments in the U.S. and Europe grapple with the tension between national AI supremacy and local environmental protections, the industry's ability to move from a model of extraction to one of community partnership will determine how quickly these massive projects can actually be built.
-
US Congress summons Nvidia CEO Jensen Huang for closed-door briefing on federal spy probe
The global dominance of Nvidia’s AI hardware has transformed the company into a focal point for international espionage and geopolitical tension. This reality has now culminated in a direct confrontation between the chipmaker's leadership and the United States government, as lawmakers seek to determine if the company's business relationships have created a vulnerability in national security.
Republican members of the U.S. House of Representatives have formally requested that Nvidia CEO Jensen Huang appear for a joint, closed-door briefing on November 18, 2026. The request, signed by Representative Brian Mast of the House Committee on Foreign Affairs, Representative John Moolenaar of the House Select Committee on the Chinese Communist Party, and Representative Rick Crawford of the House Permanent Select Committee on Intelligence, specifically targets Nvidia's security protocols against foreign espionage. The lawmakers have asked Huang to appear in room 2172 of the Rayburn House Office Building, giving him until November 11 to provide an alternative date if the proposed window is not feasible.
The urgency of the briefing stems from reports that Jensen Huang developed a close relationship with Alice Huang, an executive at the Singapore-based data center firm Megaspeed. According to reports, U.S. intelligence agencies have placed Alice Huang under surveillance, alleging that she is a People's Republic of China (PRC) intelligence asset who has funneled sensitive information to Beijing. While the specific nature of the information shared remains unverified, the lawmakers argue that the potential for PRC infiltration of cutting-edge U.S. technology puts national security at significant risk.
Hardware Security and Export Controls
Nvidia has pushed back against these allegations, maintaining that the interactions between the CEO and the Megaspeed executive were normal and not unusual. A company spokesperson noted that Jensen Huang speaks with thousands of customers annually and characterized Megaspeed as a regional cloud customer. According to the company, these customers receive ordinary business information rather than sensitive trade secrets or data relevant to national security. Nvidia further asserted that it is confident the U.S. government would have notified them if any specific individual was considered a threat to American industry.
Despite these assurances, the inquiry is underscored by evidence that advanced hardware is continuing to reach restricted markets. Reports indicate that 32 of Nvidia’s B300 GPUs—chips currently on the U.S. export control list—were discovered deployed at China Telecom. This suggests a persistent leak in the supply chain, potentially aided by smuggling loopholes where server chips equipped with High Bandwidth Memory (HBM) from manufacturers like Samsung and SK hynix are rerouted through Hong Kong and then repurposed for Chinese applications.
This hardware leak has intensified the friction between Nvidia and certain lawmakers, particularly Representative Brian Mast. Mast has frequently clashed with the company over export control policies and previously accused the firm of fighting to sell millions of chips to Chinese entities. This tension manifested in the AI Overwatch Act, a bill designed to introduce arms-style congressional oversight over AI chip sales and provide the power to block the sale of Blackwell or more advanced chips. While a version of this measure has advanced in the Senate for inclusion in an annual defense package, it lacks the joint resolution power of the original bill.
The timing of this summons is particularly volatile, as it follows midterm elections that will place both the House and Senate under Democratic control in January. While the current request is driven by Republicans, incoming Democratic leadership has signaled a desire to increase scrutiny of Silicon Valley's corporate behavior. This suggests that regardless of the outcome of the November 18 briefing, Nvidia will face a sustained period of regulatory and political pressure.
The intersection of high-end silicon and international intelligence indicates that AI infrastructure is no longer just a commercial product, but a primary instrument of state power. As the U.S. continues to tighten restrictions on the flow of advanced semiconductors, the scrutiny of Nvidia's executive dealings suggests that the government is shifting its focus from merely monitoring shipments to auditing the personal and professional networks of the industry's most powerful figures.
