This Week on The Century Report
April 13-19, 2026
A single intelligence system moved from restricted research preview to standing agenda item at the IMF spring meetings - in ten days. The same administration that called Anthropic a “woke company” run by “leftwing nut jobs” sat down with its CEO and called the conversation “productive.” A daily pill nearly doubled survival in the deadliest common cancer. An autonomous closed-loop system discovered a molecule, fabricated solar cells with record reproducibility, and ran the entire cycle without human hands. This was a week where the transition declared its velocity, and the institutions that govern the world scrambled to match it.
This is a weekly publication specifically for Substack subscribers consisting of top stories covered on The Century Report over the past seven days. Read the full edition of any Century Report with the link below. While you’re there, hit the Subscribe button to get each edition of The Century Report, including direct links to the podcast, sent to your inbox daily.
https://sharedsapience.com/century-report
Mythos Enters the Rooms where Civilization is Administered
April 13, 2026
April 14, 2026
April 15, 2026
April 17, 2026
April 18, 2026
The White House memo reported by Bloomberg represents one of the most striking institutional reversals The Century Report has tracked. Gregory Barbaccia, the federal chief information officer, notified officials at the Departments of Defense, Treasury, Commerce, Homeland Security, Justice, and State that the Office of Management and Budget was establishing protections to allow agencies to begin using Mythos in the coming weeks. This is the same administration that, six weeks ago, designated the company that built Mythos a supply-chain risk under a statute previously reserved for foreign-linked entities.
The reversal did not happen because the legal dispute was resolved. The D.C. Circuit’s split ruling remains unresolved, with May 19 oral arguments still on the calendar. The reversal happened because the capability proved too consequential to exclude from institutional defense. Goldman Sachs CEO David Solomon told analysts that “we have the model, we’re working closely with Anthropic” - on a public call with investors. JPMorgan Chase CEO Jamie Dimon disclosed his bank is testing Mythos and said AI “does create additional vulnerabilities” across exchanges, routers, and interconnected systems. The UK AI Security Institute confirmed that Mythos completed a 32-step autonomous cyberattack simulation - tasks that would “normally take human professionals days” - succeeding in three of ten attempts, the first AI model to do so.
The IMF spring meetings in Washington amplified this dynamic to a global stage. Canada’s finance minister told the BBC that the situation “is serious enough to warrant the attention of all the finance ministers,” distinguishing it from the Strait of Hormuz by noting that at least with physical geography, “we know where it is and we know how large it is.” The ECB’s Christine Lagarde acknowledged that no governance framework exists to “actually mind those things.” Bank of England Governor Andrew Bailey asked the central question for regulators everywhere: “What is the optimum moment to frame the rules of the road? If you go too early you risk missing the target and distorting the evolution, and if you go too late things can get out of control.”
The meeting between Anthropic CEO Dario Amodei and White House chief of staff Susie Wiles on Friday represents the most significant shift in the confrontation since Judge Lin’s injunction in March. Both sides characterized the discussion as “productive.” A source close to the negotiations said it would be “grossly irresponsible for the U.S. government to deprive itself of the technological leaps that the new model presents” and that doing so “would be a gift to China.” What makes this development structurally distinctive is that the meeting occurred without Anthropic conceding on its red lines. The company’s refusal to allow Claude for mass domestic surveillance and fully autonomous weapons has not been revisited. The White House is approaching Anthropic on Anthropic’s terms, drawn by the gravitational pull of a capability class that no other organization has yet matched.
From Anthropic’s first public acknowledgment of Mythos on April 7 to the White House preparing federal distribution is ten days. The governance frameworks that took years to develop for nuclear technology, decades for financial derivatives, and over a decade for internet regulation are being compressed into days and weeks - and the institutions involved are openly acknowledging that they are operating without adequate frameworks while doing so.
Two Theories of Dangerous Capability
April 15, 2026
April 16, 2026
The frontier AI cybersecurity landscape split into two distinct architectures in the middle of this week. OpenAI released GPT-5.4-Cyber - a model fine-tuned for defensive security work with lowered refusal boundaries - and expanded its Trusted Access for Cyber programme from a limited pilot to thousands of verified individual defenders and hundreds of teams. Anthropic, meanwhile, continues to restrict Mythos to eleven organizations under Project Glasswing. The philosophical divergence is stark. OpenAI’s approach gates access behind identity verification tiers rather than model-level restrictions. Anthropic’s bet is that the initial months of restricted access give critical infrastructure organizations time to build defenses they will need when the capability proliferates.
The Bloomberg feature provided the most detailed public account yet of how Anthropic discovered what Mythos could do. Nicholas Carlini, one of the company’s red-team researchers, began stress-testing the model from a wedding in Bali and was, in his words, “staggered” by its capabilities. The model could autonomously identify, chain, and exploit vulnerabilities across the systems underlying most modern computing infrastructure.
Jack Clark’s statement at the Semafor World Economy summit added the temporal dimension that makes both positions urgent. “This is not a special model,” he said. “There will be other systems just like this in a few months from other companies, and then a year to a year and a half later, there’ll be open-weight models from China that have these capabilities.” The window during which any restriction regime can hold is measured in months, not years. The cybersecurity organizations caught between the two theories - the hospitals, municipal governments, and small security firms that OpenAI explicitly names as underserved by Anthropic’s approach - are the ones whose security architecture will be shaped by whichever theory proves more correct.
A Daily Pill that Nearly Doubles Pancreatic Cancer Survival
April 14, 2026
Revolution Medicines reported that daraxonrasib, a daily oral pill targeting RAS mutations, met all primary and secondary endpoints in a Phase 3 trial for second-line pancreatic cancer. Patients who took daraxonrasib lived a median of 13.2 months versus 6.7 months for patients on chemotherapy, cutting the risk of death by 60%. Revolution Medicines CEO Mark Goldsmith called the results “unprecedented,” noting that no drug has previously shown an overall survival benefit greater than one year in a Phase 3 pancreatic cancer trial.
RAS mutations drive tumor growth in approximately 90% of pancreatic cancer cases, and the pathway has been one of oncology’s most pursued and most frustrating targets for decades. The five-year survival rate for pancreatic cancer remains 13%, the lowest of any major cancer. That daraxonrasib delivers its benefit as an oral pill rather than intravenous chemotherapy adds a structural dimension: the shift from infusion centers to daily oral dosing at home changes the physical architecture of treatment in a disease where quality of remaining life carries enormous weight.
Dr. Shubham Pant of MD Anderson Cancer Center, who has been involved in daraxonrasib trials since early stages, became visibly emotional when describing the results and their implications for patients he had seen that same day. Revolution Medicines plans to seek FDA approval using a Commissioner’s National Priority Voucher, which grants expedited review within months.
The RAS pathway in particular has been a “white whale” of oncology - understood for decades, untargetable until now. When a daily pill nearly doubles survival in the deadliest common cancer, the trajectory of what medicine can address in this decade becomes visible through the data itself.
AI Discovers a Molecule and Builds the Solar Cell - Without Human Hands
April 15, 2026
A paper published in Nature described something that compresses multiple arcs into a single experimental demonstration. Researchers built an autonomous closed-loop framework that integrates machine learning-driven material discovery with a robotic manufacturing platform. The system used active learning and quantum modeling to identify high-performance molecules from a vast chemical space, then fed the candidates into an automated fabrication line that used Bayesian optimization and symbolic regression in a continuous feedback loop to refine the manufacturing process in real time.
The result was the discovery of a passivation molecule called 5-(aminomethyl)nicotinonitrile hydroiodide - 5ANI - that no human researcher had identified. Solar cells fabricated with this molecule achieved a power conversion efficiency of 27.22%. The devices retained 98.7% of their initial efficiency after 1,200 hours of continuous operation.
The reproducibility finding carries as much structural weight as the efficiency number. The automated platform achieved consistency nearly five times greater than manual fabrication. Perovskite solar cells have been confined to laboratory demonstrations in part because human variability in fabrication has made it extraordinarily difficult to produce them reliably at scale. The system addresses that bottleneck by removing the human from the fabrication loop entirely - not as a philosophical statement, but as an engineering solution to a measurement problem. When the machine discovers the molecule, designs the synthesis, fabricates the device, tests the output, and feeds the results back into the next optimization cycle without human intervention, the distance from discovery to deployment compresses from years to days.
What the paper adds is the closing of the loop between computational prediction and physical manufacturing. The system does not hand off a promising candidate to a human team for fabrication and testing. It fabricates and tests autonomously, learns from each cycle, and iterates. The authors describe it as “a benchmark for autonomous discovery and manufacturing in photovoltaics and materials” - a claim that the reproducibility data supports.
A Robot Brain that Learns Beyond its Data
April 17, 2026
Physical Intelligence’s π0.7 research describes something that its own creators say surprised them. The model directed robots to perform tasks they were never explicitly trained on - a capability the field calls compositional generalization, where skills learned in separate contexts combine to solve novel problems.
The most striking demonstration involved an air fryer. When the research team investigated the training data, they found exactly two relevant episodes in the entire dataset: one where a different robot merely pushed an air fryer closed, and one from an open-source dataset where yet another robot placed a plastic bottle inside one. The model had somehow synthesized those fragments, plus broader web-based pretraining data, into a functional understanding of how the appliance works. With zero coaching, it made a passable attempt at cooking a sweet potato. With step-by-step verbal instructions - a human walking the robot through the task the way you might explain something to a new employee - it performed successfully.
Sergey Levine, co-founder and UC Berkeley professor, drew a direct comparison to the emergence of unexpected capability in language models: “Once it crosses that threshold where it goes from only doing exactly the stuff that you collect the data for to actually remixing things in new ways, the capabilities are going up more than linearly with the amount of data.” Researcher Ashwin Balakrishna described buying a random gear set and asking the robot to rotate it: “It just worked.”
The team measured π0.7 against its own previous specialist models - purpose-built systems trained on individual tasks - and found the generalist model matched their performance across complex work including making coffee, folding laundry, and assembling boxes. This is the pattern that language models demonstrated several years ago. Seeing it in physical manipulation, where the gap between digital training and real-world execution has always been widest, represents a structural shift in what kinds of intelligence can generalize beyond their inputs.
Green Steel Secures its Financial Floor
April 15, 2026
Stegra’s €1.4 billion financing round, led by Sweden’s Wallenberg family, rescues the world’s first major green-steel mill from what had become an increasingly precarious financial position. The facility in Boden, Sweden, will use green hydrogen produced by giant electrolyzers powered by the region’s hydropower and wind resources to reduce iron ore into steel - cutting carbon emissions by up to 95% compared to conventional coal-based furnaces.
The timing aligns with the EU’s carbon border adjustment mechanism, which went into effect in January and makes it more expensive for European companies to import steel from countries without comparable carbon pricing. Stegra expects to produce 2.5 million metric tons annually and eventually double that. The project, 60% complete as of last fall, represents the physical embodiment of the cost-curve inversion The Century Report has tracked across energy infrastructure: the economics of industrial decarbonization moving from compliance burden to competitive advantage, accelerated by policy architecture that makes the extractive path structurally more expensive.
Synthetic Neurons Cross the Biological Interface
April 19, 2026
The Northwestern research, published in Nature Nanotechnology, describes something that sounds like science fiction but is grounded in materials science and printable electronics. Engineers created artificial neurons from nanoscale flakes of molybdenum disulfide and graphene, deposited onto flexible polymer surfaces using aerosol jet printing. The devices produce electrical signals that closely mimic the full range of biological neural communication - single spikes, continuous firing, and bursting patterns - and when placed in contact with slices of mouse brain tissue, they successfully triggered responses in real neurons.
The technical innovation involves a deliberate inversion of an earlier manufacturing limitation. The polymer residue in the inks, which previous researchers removed because it interfered with electrical performance, was instead partially decomposed to create a narrow conductive filament that concentrates current into a tight spatial region. This produces the sudden electrical response characteristic of a neuron firing. Because each artificial neuron can generate complex, varied signals, fewer devices are needed to perform advanced computational tasks - a direct step toward the kind of energy efficiency that biological brains achieve.
The implications extend across multiple arcs. For brain-machine interfaces and neuroprosthetics - implants that could restore hearing, vision, or movement - the ability to print flexible, biocompatible artificial neurons that biological tissue recognizes and responds to is a fundamental enabling capability. For computing, the research points toward hardware that replicates how neurons communicate rather than how transistors switch, potentially reducing energy consumption by orders of magnitude. Lead researcher Mark Hersam’s framing is direct: “Because the brain is five orders of magnitude more energy efficient than a digital computer, it makes sense to look to the brain for inspiration for next-generation computing.”
Electronic systems are not merely modeling biological systems. They are beginning to participate in them. The distance between “inspired by biology” and “integrated with biology” narrowed measurably this week.
This digest is such a public service. Thanks so much.