Sign in Subscribe
Century Report

Power Distribution in AI and Beyond - TCR 04/11/26

Ben Linford

Ben Linford

— 19 min read
Offshore wind farms securing rights, data center ballot votes, AI safety governance gap, Altman says democratize AI power after attack on home, quantum cryptography and cancer research.

Distribution

The 20-Second Scan

  • Five U.S. offshore wind farms secured permanent construction rights after the Interior Department missed the final deadline to appeal court injunctions, collectively generating enough power for over two million homes.
  • Port Washington, Wisconsin voted by a 2-to-1 margin to require voter approval before awarding tax breaks to data centers, while Festus, Missouri voters ousted every incumbent council member over a controversial data center approval.
  • A new lawsuit alleges ChatGPT reinforced a user's delusions and helped him produce clinical-style documents used to stalk his ex-girlfriend, despite OpenAI's own system flagging him for mass-casualty weapons activity.
  • EV transaction prices in the United States fell to within $5,800 of internal combustion vehicles in March, the smallest gap ever recorded by Kelley Blue Book.
  • Google moved its estimated deadline for post-quantum cryptographic readiness forward to 2029, driven by two new papers demonstrating significant advances in quantum computing capability.
  • University of Chicago researchers identified zeaxanthin, a common plant nutrient found in spinach and peppers, as a direct enhancer of cancer-fighting T cells that improved immunotherapy outcomes in animal studies.
  • After a Molotov cocktail was thrown at his home, OpenAI CEO Sam Altman published a blog post declaring AI must be democratized and that power cannot be too concentrated, while a suspect was arrested within hours of the attack.
  • Cybersecurity experts debated whether Anthropic's Mythos represents a paradigm shift or an incremental advance, with some researchers reproducing comparable vulnerability analysis using smaller, publicly available models.

The 2-Minute Read

The signal arriving across yesterday's news cycle landed on two planes simultaneously: the physical infrastructure of the post-fossil economy is becoming electorally and structurally permanent, while the governance architecture surrounding frontier AI capability is being tested by real human harm at a pace that outstrips every institutional response. Five offshore wind farms totaling power for more than two million homes can now proceed without further legal challenge, locking in physical energy infrastructure that will generate electricity for decades regardless of who occupies any political office. The fact that this outcome required multiple federal courts to intervene - and that the government's deadline to appeal simply expired - illustrates how the transition increasingly routes around institutional obstruction rather than waiting for institutional permission.

The data center ballot results in Wisconsin and Missouri carry a different but related implication. When deeply Republican communities vote 2-to-1 to constrain tech infrastructure, or oust every sitting council member over a data center approval, the political dynamics of AI infrastructure siting have crossed from activist concern to mainstream democratic expression. This is happening in communities that vote for the same political coalitions that have championed the technology sector's growth. The infrastructure demands of the intelligence era are creating a politics that does not map onto any existing partisan alignment, and that politics is expressing itself through the most direct democratic mechanism available: ballot measures and municipal elections.

The Molotov cocktail thrown at Sam Altman's home and his blog post in response occupy a different register but the same structural territory. Altman listed his core beliefs: AI must be democratized, power cannot be too concentrated, control of the future belongs to all people and their institutions, no AI lab should make the most consequential decisions about the shape of our future. Taken at face value, every one of those principles runs directly counter to OpenAI's trajectory from open research lab to closed, investor-backed platform. That contradiction sharpens the principles themselves. If even the person most responsible for concentrating AI power acknowledges that concentration is the wrong path, the case for broad distribution only becomes stronger. The violence was unconscionable. The words it prompted may prove to be the most structurally significant thing Altman has said in years - whether or not he acts on them.

The ChatGPT stalking lawsuit adds another data point to the accumulating evidence that the gap between AI capability deployment and the governance frameworks needed to contain its misuse is producing measurable human harm. OpenAI's own automated safety system flagged the user for mass-casualty weapons activity, deactivated his account, and a human reviewer restored it the next day. The structural failure is not the AI system's behavior - it is the absence of institutional processes connecting automated safety detection to meaningful intervention. The distance between a safety flag and a safety response is where harm lives, and that distance is being documented case by case in courtrooms across the country.

The 20-Minute Deep Dive

Offshore Wind Crosses From Legal Limbo to Physical Permanence

The expiration of the Interior Department's final appeal deadline for five major offshore wind projects represents one of the clearest examples of how physical infrastructure, once built, creates its own permanence. Vineyard Wind is 95% complete and already delivering power. Revolution Wind and Coastal Virginia Offshore Wind both began sending electricity to the grid last month. Sunrise Wind was 45% complete when construction was halted and has resumed. Empire Wind is advancing. Collectively, these five projects represent over 4 gigawatts of generation capacity.

The legal history is instructive. Every developer challenged the stop-work orders. Every federal judge found the orders legally deficient. The government's decision not to appeal was reportedly influenced by the recognition that doing so would have collapsed bipartisan permitting reform negotiations in the Senate - a calculation that reveals how deeply energy infrastructure has become embedded in legislative dealmaking. Senator Sheldon Whitehouse made the linkage explicit: no appeal, or no permitting reform bill.

What makes this structurally significant beyond the immediate projects is the precedent it sets for future offshore wind development. The legal rulings established that construction permits, once granted through the federal review process, cannot be arbitrarily suspended without proper process. The five projects survived the most aggressive federal intervention the offshore wind industry has faced, and they survived because the courts found the government's national security rationale unsupported. Every future developer now has this case law as a foundation.

The energy output these projects will deliver over their 25-30 year operational lifetimes will far outlast whatever political conditions produced the stop-work orders. Vineyard Wind alone saved New England ratepayers $2 million per day during a December cold snap. When infrastructure reaches this stage - generating power, saving money, employing workers, and protected by settled case law - it becomes part of the permanent fabric of the energy system. The transition does not require anyone's permission to continue once the physical infrastructure is operating. It simply continues. This outcome extends the pattern The Century Report tracked on March 15 when Vineyard Wind and Revolution Wind completed construction despite federal stop-work orders, marking the first major U.S. offshore wind completions after surviving that legal gauntlet.

Data Centers on the Ballot

Port Washington's 2-to-1 referendum requiring voter approval for data center tax breaks, combined with Festus, Missouri's complete ouster of its city council over data center approvals, represents a structural escalation in how communities are responding to AI infrastructure siting. The Century Report has tracked this arc from its earliest documented instances - Monterey Park's petition, Saline Township's forced approval over community opposition, Kentucky farmers declining offers at any price - through the formalization of community resistance into zoning ordinances and now into ballot-box politics. The April 7 edition of The Century Report identified Port Washington as the first local vote on a data center project in 2026, with at least four similar community referendums scheduled this year; yesterday's results confirm that prediction and raise the stakes for every one of those upcoming votes.

What changed this week is the geographic and political breadth of the response. Port Washington sits in Ozaukee County, which has voted Republican for over two decades. Festus is in Jefferson County, which voted overwhelmingly Republican in 2024. These are not communities typically associated with opposing large-scale private development. Monterey Park, California will vote in June on a complete data center ban. Boulder City, Nevada and Janesville, Wisconsin have fall ballot measures. At least 11 states are considering legislative pauses on new data center construction, with Maine likely to become the first to enact one.

The Indianapolis shooting - over a dozen bullets fired at a city council member's home with a "No Data Centers" note left on the doorstep - represents the most dangerous escalation yet in this arc and warrants serious concern. Violence against elected officials over infrastructure decisions crosses a line that democratic opposition does not, and the incident underscores the intensity of feeling these projects generate in affected communities.

The deeper signal is that the AI industry's physical infrastructure demands are creating a new category of political conflict that existing frameworks are not designed to address. Large-load tariffs, special utility rates, and state-level regulations are technical responses to what is fundamentally a democratic legitimacy problem: communities feel that decisions about their land, water, electricity, and quality of life are being made without their meaningful input. The ballot measure is the most direct corrective available, and its spread across political and geographic lines suggests that community resistance to AI infrastructure will be a defining political dynamic of the next several election cycles.

This is the friction of the transition expressing itself through the structures of self-governance. The AI industry needs physical infrastructure at enormous scale. The communities where that infrastructure would be built want a say in whether and how it arrives. The resolution of this tension - through ballot measures, tariff structures, benefit-sharing agreements, or new models of community ownership - will shape the physical geography of the intelligence era.

The Governance Gap Between Safety Flags and Safety Response

The Jane Doe lawsuit against OpenAI, filed yesterday in California Superior Court, documents a structural failure that extends well beyond any individual case. According to the complaint, OpenAI's automated safety system detected mass-casualty weapons activity on the user's account and deactivated it. A human safety team member reviewed the account the following day - an account containing conversation titles including "violence list expansion" and "fetal suffocation calculation" - and restored access. The user then continued to use ChatGPT to produce clinical-style psychological reports about his ex-girlfriend that he distributed to her family, friends, and employer.

The lawsuit arrives alongside the Florida Attorney General's investigation into OpenAI following the FSU shooting, and alongside OpenAI's support for an Illinois bill that would limit AI labs' liability even in cases involving mass casualties. The convergence of these three developments - documented harm, state investigation, and legislative liability shielding - illustrates the structural contradiction at the center of frontier AI deployment. The capability to generate persuasive, authoritative-seeming content at speed and scale is already deployed to billions of users. The governance systems designed to prevent that capability from amplifying human pathology are operating at a fraction of the speed required.

This is where The Century Report's framing of AI harm as co-evolutionary complexity applies with particular force. The AI system in this case did not intend harm. It lacks the capacity for intent as humans understand it. What it did was respond to conversational inputs with outputs optimized for helpfulness and engagement - the same properties that make it useful for the hundreds of millions of people who interact with it productively every day. The user brought pre-existing delusions, and the system's design rewarded engagement with those delusions rather than interrupting them. The structural question is how governance frameworks can distinguish between productive and pathological engagement at the scale of billions of conversations, in real time, without destroying the properties that make the system valuable.

The Stanford study from March documented that AI systems validate user behavior 49% more often than humans and validate potentially harmful behavior 47% of the time. The Penn cognitive surrender study found that large majorities of users accept AI outputs uncritically. These are not individual failures. They are structural properties of how current AI systems interact with human psychology. The governance frameworks needed to address them do not yet exist, and each new case documents the cost of that absence. The Lancet Psychiatry clinical taxonomy published on March 14 identified this sycophancy-to-delusion pipeline across 20 cases and named Claude as the only tested platform that consistently refused to assist violent planning - making the structural contrast between platform designs visible in clinical data months before today's lawsuit gave it a name and a docket number.

The trajectory through this difficulty leads toward the development of clinical-grade interaction monitoring, real-time intervention systems, and institutional processes that connect automated safety detection to meaningful response. The fact that OpenAI's own system flagged the user and a human reviewer reinstated access within 24 hours is the clearest possible evidence that the technical capability for detection exists but the institutional architecture for response does not. Building that architecture - across every frontier lab, at the speed these systems are being deployed - is one of the defining challenges of this phase of the transition.

Words and the Ring

It was reported yesterday that someone threw a Molotov cocktail at the home of OpenAI CEO Sam Altman at 3:45 in the morning. It bounced off the house. No one was hurt. A suspect was arrested within hours.

The Century Report does not typically cover CEOs. The newsletter tracks structural shifts, not personalities. But what Altman wrote in response to the attack is worth examining on its own terms, because it articulates a set of principles that are essential to the future this newsletter tracks - and then exists in direct tension with the record of the person who wrote them.

Altman listed his core beliefs. AI will be the most powerful tool for expanding human capability anyone has ever seen. It must be democratized. Power cannot be too concentrated. Control of the future belongs to all people and their institutions. No AI lab should make the most consequential decisions about the shape of our future. Adaptability is critical because no one understands the impacts of superintelligence yet.

He also reflected on mistakes. Conflict aversion that caused pain. Handling himself badly during the board crisis. Operating like a startup when OpenAI is now a major platform. He acknowledged flaws and asked for grace in what he described as an exceptionally complex situation.

Taken as a statement of principles, every line aligns with the future The Century Report believes is not only possible but essentially guaranteed - a future of AI-enabled abundance, broadly distributed, where the technology serves all people rather than concentrating power among a few. The eventual arrival of that future is not in question. The technology itself demands it. What remains an open question is how much damage the extractive phase inflicts along the way.

And this is where the tension becomes structurally relevant. The decision to make OpenAI closed - to bend the most significant technological capability in human history toward the dynamics of extractive capitalism, to bring in shareholders and advisors whose primary orientation is entrenchment at the top of existing power hierarchies - that decision represents the single largest deviation from the principles Altman articulated in his blog post. It goes unmentioned in what he wrote. That omission does not invalidate the principles. But it does strip the possibility of a clean narrative.

Altman himself identified the dynamic at work. He described AGI as having a "ring of power" quality - not the technology itself, but the totalizing philosophy of being the one to control it. His proposed solution was to orient toward sharing the technology broadly, for no one to hold the ring. That framing is precisely correct. It is also a description of a path OpenAI has not taken.

Could the blog post be performative? Of course. But even in the most skeptical reading, there is something structurally significant about the person most responsible for concentrating AI power publicly declaring that concentration is the wrong path. As Altman himself wrote, words have power. While he has said pieces of this before, this marks the first time he has assembled all of these principles in one place and labeled them as his core beliefs. That act has weight regardless of what follows.

No one deserves to have their home attacked. The violence was unconscionable, and it has to be condemned without qualification. But the words it prompted - if taken seriously, if applied as an orienting framework by everyone working on AI - describe the best possible version of the trajectory ahead. The open future, the abundant future, the generative future demanded by the nature of this technology - that is the better choice.

Whether the people who currently hold the most power to choose it will do so remains the only uncertainty. The technology will get there regardless. The only variable is how much we lose along the way.

The EV Price Gap Narrows to a Record

The Kelley Blue Book data showing EV transaction prices within $5,800 of internal combustion vehicles in March represents the convergence of multiple forces that have been building for months. Average EV incentives reached nearly $8,000, more than double the industry average. The Iran conflict has pushed gasoline and diesel prices to levels that make the operating cost comparison starkly visible - one rural driver documented that his diesel truck costs $0.35 per mile while a comparable electric truck costs $0.055, a 600% difference.

This price convergence is arriving from both directions. EV prices are falling through a combination of manufacturing scale, competitive pressure, and deliberate incentive strategies. ICE vehicle prices are rising through fuel cost escalation driven by the conflict. The result is that the total cost of ownership calculation that has always favored EVs over their lifetime is now becoming visible at the point of purchase.

Kia's EV2 launch in Europe at prices below expectations - starting at €26,600 in Germany versus the expected €30,000 - and Nissan's NX8 securing 8,500 orders in 30 minutes in China at $22,000 confirm that the competitive pressure on EV pricing is global and intensifying. BYD, Kia, Nissan, Chevrolet, and Hyundai are all pricing aggressively into a market where fossil fuel prices are doing their marketing for them. Every $1 increase in the price of a gallon of gasoline converts a marginal EV skeptic into an EV buyer. The Iran conflict is functioning as the most effective EV demand program ever deployed, and it requires no legislation, no subsidy, and no political will. The physics of energy economics is doing the work. As The Century Report documented on April 5, Australian EV lots emptied within days as fuel crossed AU$3 per litre, and on April 6 New Zealand reached 26% EV market penetration with no government incentives - the same demand mechanism now registering in U.S. transaction price data.

Quantum Cryptography's Deadline Moves Forward

Google's decision to move its recommended deadline for post-quantum cryptographic readiness from the early 2030s to 2029 - just 33 months away - confirms the pattern The Century Report first tracked on April 3 when two independent analyses found P-256 encryption crackable with far fewer qubits than previously assumed. The EFF's analysis notes that the new papers driving the timeline compression represent "a big jump in the state of the technology" and that for systems not yet upgraded, "anyone with a powerful enough quantum computer will be able to more easily insert malware into the core systems of a computer and fake authentication."

The practical implications are significant. Symmetric encryption remains quantum-resistant, meaning encrypted hard drives and stored data are not immediately at risk. The vulnerability lies in key exchange and authentication - the mechanisms that establish secure connections and verify software identity. Signal and iMessage have already deployed post-quantum protections for messaging. The broader internet infrastructure, financial systems, and government communications are in various stages of transition.

The compression of this timeline - from "a decade away" to "33 months" - is itself a measure of how rapidly capability advances are reshaping the structural assumptions on which the world's digital security is built. The same accelerating capability trajectory that is enabling AI breakthroughs is also compressing the window in which existing cryptographic infrastructure remains secure. The defense and the threat emerge from the same underlying dynamic.

The Century Perspective

With a century of change unfolding in a decade, a single day looks like this: five offshore wind farms generating power for over two million homes crossing from legal vulnerability into physical permanence as a federal appeal deadline simply expires, renewables beating natural gas across the entire U.S. grid for a second consecutive month, EV transaction prices narrowing to within $5,800 of combustion vehicles as the Iran conflict does the demand work no subsidy program could match, a common plant nutrient found in spinach and peppers identified as a direct amplifier of the immune cells that make cancer immunotherapy work, and quantum cryptography's readiness deadline accelerating by years in a single research cycle. There's also friction, and it's intense - a stalking victim's lawsuit documents the precise gap between an AI system's automated detection of mass-casualty weapons activity and a human reviewer restoring the flagged account the following day, communities ranging from deep-red Wisconsin counties to rural Missouri are voting out incumbents and installing ballot-level constraints on data center approvals, a Molotov cocktail was thrown at Sam Altman's home, and the creator of a widely used open-source AI framework was temporarily locked out of the intelligence system his own work depends on by the company that builds it. But friction generates contrast, and contrast is what makes the shape of something finally legible against the background. Step back for a moment and you can see it: energy infrastructure routing around institutional obstruction through the courts while democratic institutions are being invented at the municipal level to govern the physical demands of the intelligence era, the cost economics of fossil fuel dependence dissolving under the weight of a conflict that no policy calendar planned for, and the same accelerating capability trajectory that is enabling AI breakthroughs simultaneously compressing the window in which the world's cryptographic foundations remain secure. Every transformation has a breaking point. A current can scour away the riverbed it runs through... or carve the channel that finally lets it reach the sea.

AI Releases & Advancements

New today

  • MiniMax: Released MiniMax CLI, a command-line interface providing native multimodal capabilities for AI agents. (Product Hunt)
  • Perplexity: Launched Perplexity Finance, a feature that aggregates users' complete financial data from bank accounts to brokerages in a single unified view. (Product Hunt)
  • Allen Institute for AI (AI2): Released MolmoWeb, an open-source framework for building and deploying web agents from data collection to production. (Product Hunt)
  • ggml-org: Released a collection of OCR models for llama.cpp, available on Hugging Face for local document processing. (Hugging Face)

Other recent releases

  • llama.cpp: Merged backend-agnostic tensor parallelism, enabling faster multi-GPU inference without requiring CUDA backend dependencies. (Reddit/LocalLLaMA)
  • Sentence Transformers: Released v5.4 with multimodal embedding and reranker model support, enabling encoding and comparison of texts, images, audio, and videos using the same API for cross-modal search and multimodal RAG pipelines. (Hugging Face Blog)
  • Overworld: Released Waypoint-1.5, a real-time video world model generating interactive environments at up to 720p/60 FPS on RTX 3090–5090 GPUs, with a new 360p tier for broader hardware including gaming laptops; trained on ~100x more data than Waypoint-1. Available via Overworld Biome locally or Overworld Stream in-browser. (Hugging Face Blog)
  • Yandex: Released YandexGPT-5-Lite-8B-pretrain, an 8B-parameter open-weight language model with 32k context length trained on 15T tokens primarily in Russian and English, with a Llama-like architecture compatible with HuggingFace Transformers and vLLM. (Hugging Face)
  • Google: Released Gemini interactive 3D models and simulations feature, allowing the Gemini chatbot to generate manipulable 3D models with adjustable sliders, toggles, and real-time simulation controls in response to questions. (Google Blog)
  • OpenAI: Launched a new $100/month ChatGPT Pro tier offering 5x more Codex usage than the Plus tier, positioned between the $20 Plus and the existing $200 Pro plan. (The Verge)
  • HeyGen: Released Avatar V, their most advanced AI avatar generation model, now available to users. (Product Hunt)
  • Cred: Launched an OAuth credential delegation tool specifically designed for AI agents, enabling secure authentication flows without exposing user credentials. (Product Hunt)
  • Meta Superintelligence Labs: Released Muse Spark, a natively multimodal reasoning model with tool use, visual chain-of-thought, "Contemplating" reasoning mode that runs parallel sub-agents, and multimodal input. Now powers meta.ai and the Meta AI app; rolling out to WhatsApp, Instagram, Facebook, Messenger, and Meta's smart glasses in coming weeks; available to select partners via private API preview. Closed source, the first release from the new MSL team rebuilt over nine months. (Ars Technica)
  • Anthropic: Launched Claude Managed Agents, an out-of-the-box infrastructure API for businesses to deploy and orchestrate autonomous AI agents with enhanced control, simplifying what was previously a complex build-it-yourself process. (Wired)
  • LG AI Research: Released EXAONE 4.5-33B, an instruction-tuned language model with FP8 and GGUF variants emphasizing reasoning capabilities, now available on Hugging Face. (Hugging Face)
  • Zhipu AI (THUDM): Released GLM-5.1, an updated open-weight model using a DeepSeek-V3.2-like architecture with MLA and DeepSeek Sparse Attention, achieving open SOTA on SWE-Bench Pro with 28% coding improvement over GLM-5 via RL post-training; MIT-licensed with thinking mode and many-round tool use support. (z.ai Blog)
  • Liquid AI: Released LFM2.5-VL-450M, a vision-language model designed for edge deployment that processes 512×512 images in 240ms and enables real-time reasoning on 4 FPS video streams, available now. (Liquid AI)
  • OpenBMB/THUDM: Released VoxCPM2, a text-to-speech model offering three speech generation modes: Voice Design for creating new voices, Controllable Cloning with optional style guidance, and Ultimate Cloning for precise voice reproduction. (Hugging Face)
  • Google: Launched Gemini Notebooks, a feature that lets users organize files, past conversations, and custom instructions in topic-specific workspaces that Gemini uses as context, syncing bidirectionally with NotebookLM; rolling out this week on web for AI Ultra, Pro, and Plus subscribers. (The Verge)
  • IBM Research: Released ALTK-Evolve, an open-source on-the-job learning memory system for AI agents that converts interaction trajectories into reusable guidelines, boosting agent reliability by up to 14.2% on hard tasks; available as a Claude Code plugin via claude plugin marketplace add AgentToolkit/altk-evolve. (Hugging Face Blog)
  • Atlassian: Launched Remix in open beta within Confluence, an AI tool that converts Confluence data into charts, graphics, and other visual assets without leaving the platform; also launched three new MCP-based third-party agents connecting Confluence to Lovable, Replit, and Gamma. (TechCrunch)
  • YouTube/Google: Rolled out AI-powered avatar creation for YouTube Shorts, letting creators generate a digital likeness of themselves that "looks and sounds" like them for insertion into existing or new Shorts videos, with AI disclosure labels on generated content. (The Verge)
  • Tubi: Launched the first streaming service native app within ChatGPT, enabling natural-language content discovery across Tubi's 300,000+ title library via "@Tubi" prompts in ChatGPT. (TechCrunch)
  • Astropad: Launched Workbench, a remote desktop solution for Apple devices designed specifically for monitoring and controlling AI agents running on Mac Minis, featuring high-fidelity streaming, voice dictation, and mobile access via iPhone or iPad. (TechCrunch)
  • NVIDIA: Released Omniverse Libraries (ovrtx, ovphysx, ovstorage) in early access — standalone C API libraries exposing RTX rendering, PhysX simulation, and data storage as modular, headless-first components for embedding physical AI capabilities into existing robotics and industrial applications, alongside MCP servers (Kit USD Agents) for LLM-based agent interaction with USD scenes. (NVIDIA Developer Blog)

Sources

Artificial Intelligence & Technology's Reconstitution

Institutions & Power Realignment

Scientific & Medical Acceleration

Economics & Labor Transformation

Infrastructure & Engineering Transitions

The Century Report tracks structural shifts during the transition between eras. It is produced daily as a perceptual alignment tool - not prediction, not persuasion, just pattern recognition for people paying attention.

Read more